AI Driven Spreadsheet

AI Driven Spreadsheet — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

• 

  Elements (toolchain)

  RemObjects Elements is a toolchain for software development, comprising six programming languages: C#, Swift, Go, Java, Oxygene (a form of modern Object Pascal), and Visual Basic .NET. All languages interoperate, meaning a single project can use any combination of languages, and they can all be compiled to .NET, the JVM, native, or WebAssembly. Elements supports Microsoft Windows, all Apple Inc. platforms (including iOS, visionOS and watchOS), Android, and Linux. Elements also supports language conversion, allowing source code in one language to be rewritten in another. Elements is supported in Visual Studio, but RemObjects also makes their own IDEs, Fire (on MacOS) and Water (on Windows.) == Background == RemObjects began in 2002, creating software for Delphi, but in 2005 in response to the growth of .NET and that Delphi was targeting only native Windows, they released Oxygene (known as Chrome at the time) as a new version of Object Pascal, with more modern syntax as well as being .NET-native. Since then, five other languages have been added to the suite, as well as compiling for the web via WebAssembly and to native architectures (eg Intel 32/64 or ARM64). Elements is primarily intended for developers who want to pull together libraries and codebases written in multiple languages, including legacy codebases in older languages while modernizing either with newer syntax and features or by adding in the use of newer or more popular languages. Because of the Oxygene flavour of Object Pascal, supporting Delphi apps is a primary focus, including allowing Pascal to be compiled for other architectures or providing language features that match other prominent languages. == Approach == New versions of the Elements come out approximately every week. RemObjects names its programming languages after chemical elements, sometimes with poetic or musical spelling, rather than referring to them directly. They are: C#: Hydrogene Object Pascal: Oxygene Java: Iodine Visual Basic: Mercury Go: Gold Swift: Silver == History == The Elements compiler was first introduced with version 1.0 in 2005 under the name "Chrome", with support for only the Oxygene language on the .NET platform, primarily as a response to the then-new and not well-received Delphi .NET compiler from Embarcadero. Chrome saw updates to version 1.5 'Floorshow' and Chrome 2.0 'Joyride' over the next few years, moving in parallel with major advancements on the .NET platform for .NET 2.0 (Generics) and .NET 3.x (LINQ), respectively. With the release of version 3.0 (code-named Oxygène after the Jean-Michel Jarre album of the same name) Chrome was rebranded to Oxygene in 2008, and also shipped co-branded by Embarcadero as Delphi Prism (later just Prism) as part of RAD Studio, replacing Embarcadero's own and now-defunct Delphi.NET compiler. 2010 saw the release of Oxygene 4 ("Echoes"), the last version to focus on just a single language and platform. With Oxygene 5 in 2011 and Oxygene 6 in 2013, RemObjects introduced new platform support for Java/Android (code-name "Cooper") and then Cocoa, the Apple development platform (code-name "Toffee"). Elements 7.0 was released at the beginning of 2014, adding the second programming language, C# to the compiler, and delegating Oxygene from the product name to merely branding the Object Pascal-based language. Over the subsequent years, Elements gained support for additional languages, with Apple Swift in 2015, Java in 2017, and subsequently Google's Go and Mercury, a revitalized Visual Basic.NET. Elements also gained support for its fourth target platform, "Island", for CPU-native compilation for Windows, Linux, and WebAssembly. In addition to the chemical elements-based names for the different languages, the "Elements" concept was carried on with the introduction of dedicated development environments alchemically named Fire (for the Mac, in 2015) and Water (for Windows, in 2018). == Fire and Water (IDEs) == Fire and Water are integrated development environments developed by RemObjects Software. They are designed specifically for use with the Elements Compiler. Fire is the version developed for macOS, while Water is intended for Microsoft Windows. Both IDEs are designed to work closely with the Elements compiler and are primarily intended for developers using the RemObjects language ecosystem. They support software development across multiple platforms, including .NET, Android, iOS, macOS, Windows, Linux, and WebAssembly. The IDEs include standard development tools such as syntax highlighting, code completion, debugging, and project navigation. Build operations are managed using a custom system known as EBuild, which is part of the broader Elements toolchain. The IDEs are distributed as part of the RemObjects Elements package and are updated in coordination with the compiler itself. == In media == Oxygene has been mentioned several times by Verity Stob in their Chronicles of Delphi series, currently living at The Register.

  Read more →
• 

  Harris corner detector

  The Harris corner detector is a corner detection operator that is commonly used in computer vision algorithms to extract corners and infer features of an image. It was first introduced by Chris Harris and Mike Stephens in 1988 upon the improvement of Moravec's corner detector. Compared to its predecessor, Harris' corner detector takes the differential of the corner score into account with reference to direction directly, instead of using shifting patches for every 45 degree angles, and has been proved to be more accurate in distinguishing between edges and corners. Since then, it has been improved and adopted in many algorithms to preprocess images for subsequent applications. == Introduction == A corner is a point whose local neighborhood stands in two dominant and different edge directions. In other words, a corner can be interpreted as the junction of two edges, where an edge is a sudden change in image brightness. Corners are the important features in the image, and they are generally termed as interest points which are invariant to translation, rotation and illumination. Although corners are only a small percentage of the image, they contain the most important features in restoring image information, and they can be used to minimize the amount of processed data for motion tracking, image stitching, building 2D mosaics, stereo vision, image representation and other related computer vision areas. In order to capture the corners from the image, researchers have proposed many different corner detectors including the Kanade-Lucas-Tomasi (KLT) operator and the Harris operator which are most simple, efficient and reliable for use in corner detection. These two popular methodologies are both closely associated with and based on the local structure matrix. Compared to the Kanade-Lucas-Tomasi corner detector, the Harris corner detector provides good repeatability under changing illumination and rotation, and therefore, it is more often used in stereo matching and image database retrieval. Although there still exist drawbacks and limitations, the Harris corner detector is still an important and fundamental technique for many computer vision applications. == Development of Harris corner detection algorithm == Source: Without loss of generality, we will assume a grayscale 2-dimensional image is used. Let this image be given by I {\displaystyle I} . Consider taking an image patch ( x , y ) ∈ W {\displaystyle (x,y)\in W} (window) and shifting it by ( Δ x , Δ y ) {\displaystyle (\Delta x,\Delta y)} . The sum of squared differences (SSD) between these two patches, denoted f {\displaystyle f} , is given by: f ( Δ x , Δ y ) = ∑ ( x k , y k ) ∈ W ( I ( x k , y k ) − I ( x k + Δ x , y k + Δ y ) ) 2 {\displaystyle f(\Delta x,\Delta y)={\underset {(x_{k},y_{k})\in W}{\sum }}\left(I(x_{k},y_{k})-I(x_{k}+\Delta x,y_{k}+\Delta y)\right)^{2}} I ( x + Δ x , y + Δ y ) {\displaystyle I(x+\Delta x,y+\Delta y)} can be approximated by a Taylor expansion. Let I x {\displaystyle I_{x}} and I y {\displaystyle I_{y}} be the partial derivatives of I {\displaystyle I} , such that I ( x + Δ x , y + Δ y ) ≈ I ( x , y ) + I x ( x , y ) Δ x + I y ( x , y ) Δ y {\displaystyle I(x+\Delta x,y+\Delta y)\approx I(x,y)+I_{x}(x,y)\Delta x+I_{y}(x,y)\Delta y} This produces the approximation f ( Δ x , Δ y ) ≈ ∑ ( x , y ) ∈ W ( I x ( x , y ) Δ x + I y ( x , y ) Δ y ) 2 , {\displaystyle f(\Delta x,\Delta y)\approx {\underset {(x,y)\in W}{\sum }}\left(I_{x}(x,y)\Delta x+I_{y}(x,y)\Delta y\right)^{2},} which can be written in matrix form: f ( Δ x , Δ y ) ≈ ( Δ x Δ y ) M ( Δ x Δ y ) , {\displaystyle f(\Delta x,\Delta y)\approx {\begin{pmatrix}\Delta x&\Delta y\end{pmatrix}}M{\begin{pmatrix}\Delta x\\\Delta y\end{pmatrix}},} where M is the structure tensor, M = ∑ ( x , y ) ∈ W [ I x 2 I x I y I x I y I y 2 ] = [ ∑ ( x , y ) ∈ W I x 2 ∑ ( x , y ) ∈ W I x I y ∑ ( x , y ) ∈ W I x I y ∑ ( x , y ) ∈ W I y 2 ] {\displaystyle M={\underset {(x,y)\in W}{\sum }}{\begin{bmatrix}I_{x}^{2}&I_{x}I_{y}\\I_{x}I_{y}&I_{y}^{2}\end{bmatrix}}={\begin{bmatrix}{\underset {(x,y)\in W}{\sum }}I_{x}^{2}&{\underset {(x,y)\in W}{\sum }}I_{x}I_{y}\\{\underset {(x,y)\in W}{\sum }}I_{x}I_{y}&{\underset {(x,y)\in W}{\sum }}I_{y}^{2}\end{bmatrix}}} == Process of Harris corner detection algorithm == Commonly, Harris corner detector algorithm can be divided into five steps. Color to grayscale Spatial derivative calculation Structure tensor setup Harris response calculation Non-maximum suppression === Color to grayscale === If we use Harris corner detector in a color image, the first step is to convert it into a grayscale image, which will enhance the processing speed. The value of the gray scale pixel can be computed as a weighted sums of the values R, B and G of the color image, ∑ C ∈ { R , G , B } w C ⋅ C {\displaystyle \sum _{C\,\in \,\{R,G,B\}}w_{C}\cdot C} , where, e.g., w R = 0.299 , w G = 0.587 , w B = 1 − ( w R + w G ) = 0.114. {\displaystyle w_{R}=0.299,\ w_{G}=0.587,\ w_{B}=1-(w_{R}+w_{G})=0.114.} === Spatial derivative calculation === Next, we are going to find the derivative with respect to x and the derivative with respect to y, I x ( x , y ) {\displaystyle I_{x}(x,y)} and I y ( x , y ) {\displaystyle I_{y}(x,y)} . This can be approximated by applying Sobel operators. === Structure tensor setup === With I x ( x , y ) {\displaystyle I_{x}(x,y)} , I y ( x , y ) {\displaystyle I_{y}(x,y)} , we can construct the structure tensor M {\displaystyle M} . === Harris response calculation === For x ≪ y {\displaystyle x\ll y} , one has x ⋅ y x + y = x 1 1 + x / y ≈ x . {\displaystyle {\tfrac {x\cdot y}{x+y}}=x{\tfrac {1}{1+x/y}}\approx x.} In this step, we compute the smallest eigenvalue of the structure tensor using that approximation: λ min ≈ λ 1 λ 2 ( λ 1 + λ 2 ) = det ( M ) tr ⁡ ( M ) {\displaystyle \lambda _{\min }\approx {\frac {\lambda _{1}\lambda _{2}}{(\lambda _{1}+\lambda _{2})}}={\frac {\det(M)}{\operatorname {tr} (M)}}} with the trace t r ( M ) = m 11 + m 22 {\displaystyle \mathrm {tr} (M)=m_{11}+m_{22}} . Another commonly used Harris response calculation is shown as below, R = λ 1 λ 2 − k ( λ 1 + λ 2 ) 2 = det ( M ) − k tr ⁡ ( M ) 2 {\displaystyle R=\lambda _{1}\lambda _{2}-k(\lambda _{1}+\lambda _{2})^{2}=\det(M)-k\operatorname {tr} (M)^{2}} where k {\displaystyle k} is an empirically determined constant; k ∈ [ 0.04 , 0.06 ] {\displaystyle k\in [0.04,0.06]} . === Non-maximum suppression === In order to pick up the optimal values to indicate corners, we find the local maxima as corners within the window which is a 3 by 3 filter. == Improvement == Sources: Harris-Laplace Corner Detector Differential Morphological Decomposition Based Corner Detector Multi-scale Bilateral Structure Tensor Based Corner Detector == Applications == Image Alignment, Stitching and Registration 2D Mosaics Creation 3D Scene Modeling and Reconstruction Motion Detection Object Recognition Image Indexing and Content-based Retrieval Video Tracking

  Read more →
• 

  Dental AI

  Dental artificial intelligence (Dental AI) refers to the application of artificial intelligence (AI) and machine-learning methods to oral healthcare data. These systems can be used to find patterns or make predictions that can aid in diagnosis, treatment, patient communication, or practice management. == History and development == Research into AI for dentistry dates to the 1990s and 2000s, alongside early CAD/CAM and image-analysis work in dental radiology. Recent developments in deep learning, especially those involving computer vision, such as convolutional neural networks, trained on large image datasets, led to a rapid improvement in performance, as well as a move from prototype technology to productization suitable for use in dental chairs. Dental schools and continuing education programs started incorporating AI content in the 2020s. == Definition and core technologies == The dental AI software accomplishes this task by using various dental images and patient data. Dental images and data used by the dental AI software include bitewing and periapical X-rays, complete mouth X-rays, detailed 3D images, intraoral images, and the patient’s medical history. The dental AI software utilizes several core technologies in accomplishing its task of assisting the dentist. First, the dental AI software utilizes machine learning and deep learning using programs that can learn from examples. Such programs are referred to as convolutional neural network (CNN) and can detect cavities and identify bone changes related to gum disease. The dental AI software utilizes computer vision, which enables the AI software to identify and quantify important features in images and data, whether they are 2D images or 3D images. Natural language processing (NLP) is used for the AI software to understand written text and can automatically generate dental notes and communicate with the patient. Furthermore, the dental AI software utilizes predictive analytics to identify patients that are more prone to dental complications and can suggest the best intervals for checkups or future dental procedures. == Applications in dentistry == Reported clinical and operational applications include diagnostic assistance for caries and periodontal disease, treatment planning assistance, patient education overlays, quality assurance, curriculum assistance for dental education, and claims documentation. Systematic reviews continue to find image-based applications such as caries detection with some variability in study design and a need for prospective validation. == Academic research and clinical validation == Several peer-reviewed studies have measured the effectiveness of AI for applications such as interproximal caries detection and periodontal bone level assessment, showing improvements over unaided readings with a focus on bias within the dataset. The Dental AI Council found variability among clinicians for diagnosis and treatment planning, suggesting the use of a standard tool as an assist. == Industry adoption == Multiple vendors offer FDA-cleared chairside AI for dental imaging: Pearl — Received U.S. FDA 510(k) clearance for its real-time radiologic aid (“Second Opinion”) in 2022 (2D), with subsequent clearances including pediatric and CBCT (“Second Opinion 3D”). TIME gave “Second Opinion” a special mention on its Best Inventions of 2022 list. Overjet — FDA-cleared for bone-level quantification and detection/outline of caries and calculus (e.g., K210187), with additional clearances expanding capabilities. VideaHealth — Received an FDA 510(k) covering 30+ detections across common dental findings (K232384), including indications for patients ages 3 and up; trade coverage has described elements of this as the first pediatric dental-AI clearance. == Regulations == In the U.S., AI-enabled dental imaging software is generally reviewed via the FDA’s 510(k) pathway. The FDA maintains a public AI-Enabled Medical Devices List, which includes numerous medical-imaging AI tools (including dental). Specific dental clearances include Overjet (K210187), VideaHealth (K232384), and Pearl entries such as “Second Opinion 3D” (K243989).

  Read more →
• 

  Speculative decoding

  Speculative decoding is an inference-time optimization for autoregressive large language models (LLMs) that generates multiple tokens per decoding step instead of one. A smaller draft model proposes a sequence of candidate tokens, and the larger target model verifies them in a single forward pass through a modified rejection sampling scheme. The verification preserves the target model's original output distribution, so the technique produces the same results as standard decoding while cutting latency by roughly two to three times. The name is an analogy to speculative execution in CPU design, where a processor runs instructions along a predicted branch before the outcome is known. == Background == Standard autoregressive decoding in large language models generates one token at a time. The model computes a probability distribution over its vocabulary, samples the next token, and feeds that token back as input. For large models, this process is bottlenecked by memory bandwidth rather than arithmetic throughput: loading the model's parameters from high-bandwidth memory (HBM) to the processor takes up most of the wall-clock time at each step. Because of this, a forward pass over one token and a forward pass over several tokens in a batch take roughly the same time. Speculative decoding relies on this property. == Mechanism == The technique alternates between two phases: drafting and verification. During drafting, a fast approximation model generates a short run of K candidate tokens, typically between 3 and 12. The draft model is usually a much smaller version of the target model or a lightweight auxiliary network. During verification, the target model scores the entire draft sequence in one batched forward pass. A modified rejection sampling algorithm compares the draft and target probabilities at each position. If the target model would have been at least as likely to produce a given token, that token is accepted; the first token that fails is resampled from a corrected distribution, and everything after it is thrown out. The result is that the output distribution is the same as if each token had been generated one at a time. How many tokens get accepted per cycle depends on how well the draft model matches the target. For common words and predictable continuations the match tends to be good, so the target model can confirm several tokens at once. == History == An early precursor was blockwise parallel decoding, proposed in 2018 by Stern, Shazeer, and Uszkoreit. Their method predicted multiple future tokens through auxiliary prediction heads and validated them against the autoregressive model, but it only worked with greedy decoding and did not preserve the full sampling distribution. The modern form of the technique came from Yaniv Leviathan, Matan Kalman, and Yossi Matias at Google Research, who posted "Fast Inference from Transformers via Speculative Decoding" on arXiv in November 2022. Separately and at about the same time, Charlie Chen and colleagues at DeepMind arrived at a closely related method they called speculative sampling, published in February 2023. Both papers introduced the use of rejection sampling to guarantee that the output distribution is unchanged. Leviathan et al. showed roughly 2–3x speedup on T5-XXL (11 billion parameters); Chen et al. reported 2–2.5x on the Chinchilla model (70 billion parameters). The Leviathan et al. paper was presented as an oral at the International Conference on Machine Learning in July 2023. == Variants == SpecInfer (Miao et al., 2024) uses multiple small language models to jointly build a tree of candidate continuations rather than a single chain. The target model verifies the whole tree in parallel and keeps the longest valid path, with reported speedups of 1.5–3.5x. Medusa (Cai et al., 2024) takes a different approach by not using a separate draft model at all. Extra lightweight decoding heads are attached to the target model itself, and each one predicts a token at a different future position. The candidates are evaluated through a tree-structured attention mechanism. The authors measured 2.2–3.6x speedup. EAGLE (Li et al., 2024) performs autoregression on the target model's internal feature representations (specifically the second-to-top layer) rather than on tokens directly. On LLaMA 2 Chat 70B, this gave a 2.7–3.5x latency reduction. Later versions added dynamic draft trees (EAGLE-2) and further optimizations (EAGLE-3), reaching 3–6.5x speedup. == Adoption == By 2024, speculative decoding had become a standard part of production LLM serving. Google uses it in the AI Overviews feature of Google Search. Open-source inference frameworks such as vLLM, NVIDIA's TensorRT-LLM, and SGLang all include built-in support for speculative decoding and its variants. Apple, AWS, and Meta have also published research extending the method or deploying it at scale.

  Read more →
• 

  RCUDA

  rCUDA, which stands for Remote CUDA, is a type of middleware software framework for remote GPU virtualization. Fully compatible with the CUDA application programming interface (API), it allows the allocation of one or more CUDA-enabled GPUs to a single application. Each GPU can be part of a cluster or running inside of a virtual machine. The approach is aimed at improving performance in GPU clusters that are lacking full utilization. GPU virtualization reduces the number of GPUs needed in a cluster, and in turn, leads to a lower cost configuration – less energy, acquisition, and maintenance. The recommended distributed acceleration architecture is a high performance computing cluster with GPUs attached to only a few of the cluster nodes. When a node without a local GPU executes an application needing GPU resources, remote execution of the kernel is supported by data and code transfers between local system memory and remote GPU memory. rCUDA is designed to accommodate this client-server architecture. On one end, clients employ a library of wrappers to the high-level CUDA Runtime API, and on the other end, there is a network listening service that receives requests on a TCP port. Several nodes running different GPU-accelerated applications can concurrently make use of the whole set of accelerators installed in the cluster. The client forwards the request to one of the servers, which accesses the GPU installed in that computer and executes the request in it. Time-multiplexing the GPU, or in other words sharing it, is accomplished by spawning different server processes for each remote GPU execution request. == rCUDA v20.07 == The rCUDA middleware enables the concurrent usage of CUDA-compatible devices remotely. rCUDA employs either the InfiniBand network or the socket API for the communication between clients and servers. rCUDA can be useful in three different environments: Clusters. To reduce the number of GPUs installed in High Performance Clusters. This leads to energy savings, as well as other related savings like acquisition costs, maintenance, space, cooling, etc. Academia. In commodity networks, to offer access to a few high performance GPUs concurrently to many students. Virtual Machines. To enable the access to the CUDA facilities on the physical machine. The current version of rCUDA (v20.07) supports CUDA version 9.0, excluding graphics interoperability. rCUDA v20.07 targets the Linux OS (for 64-bit architectures) on both client and server sides. CUDA applications do not need any change in their source code in order to be executed with rCUDA.

  Read more →
• 

  History of machine translation

  Machine translation is a sub-field of computational linguistics that investigates the use of software to translate text or speech from one natural language to another. In the 1950s, machine translation became a reality in research, although references to the subject can be found as early as the 17th century. The Georgetown experiment, which involved successful fully automatic translation of more than sixty Russian sentences into English in 1954, was one of the earliest recorded projects. Researchers of the Georgetown experiment asserted their belief that machine translation would be a solved problem within a few years. In the Soviet Union, similar experiments were performed shortly after. Consequently, the success of the experiment ushered in an era of significant funding for machine translation research in the United States. The achieved progress was much slower than expected; in 1966, the ALPAC report found that ten years of research had not fulfilled the expectations of the Georgetown experiment and resulted in dramatically reduced funding. Interest grew in statistical models for machine translation, which became more common and also less expensive in the 1980s as available computational power increased. Although there exists no autonomous system of "fully automatic high quality translation of unrestricted text," there are many programs now available that are capable of providing useful output within strict constraints. Several of these programs are available online, such as Google Translate and the SYSTRAN system that powers AltaVista's BabelFish (which was replaced by Microsoft Bing translator in May 2012). == The beginning == The origins of machine translation can be traced back to the work of Al-Kindi, a 9th-century Arabic cryptographer who developed techniques for systemic language translation, including cryptanalysis, frequency analysis, and probability and statistics, which are used in modern machine translation. The idea of machine translation later appeared in the 17th century. In 1629, René Descartes proposed a universal language, with equivalent ideas in different tongues sharing one symbol. In the mid-1930s the first patents for "translating machines" were applied for by Georges Artsrouni, for an automatic bilingual dictionary using punched tape. Russian Peter Troyanskii submitted a more detailed proposal that included both the bilingual dictionary and a method for dealing with grammatical roles between languages, based on the grammatical system of Esperanto. This system was separated into three stages: stage one consisted of a native-speaking editor in the source language to organize the words into their logical forms and to exercise the syntactic functions; stage two required the machine to "translate" these forms into the target language; and stage three required a native-speaking editor in the target language to normalize this output. Troyanskii's proposal remained unknown until the late 1950s, by which time computers were well-known and utilized. == The early years == The first set of proposals for computer based machine translation was presented in 1949 by Warren Weaver, a researcher at the Rockefeller Foundation, "Translation memorandum". These proposals were based on information theory, successes in code breaking during the Second World War, and theories about the universal principles underlying natural language. A few years after Weaver submitted his proposals, research began in earnest at many universities in the United States. On 7 January 1954 the Georgetown–IBM experiment was held in New York at the head office of IBM. This was the first public demonstration of a machine translation system. The demonstration was widely reported in the newspapers and garnered public interest. The system itself, however, was no more than a "toy" system. It had only 250 words and translated 49 carefully selected Russian sentences into English – mainly in the field of chemistry. Nevertheless, it encouraged the idea that machine translation was imminent and stimulated the financing of the research, not only in the US but worldwide. Early systems used large bilingual dictionaries and hand-coded rules for fixing the word order in the final output which was eventually considered too restrictive in linguistic developments at the time. For example, generative linguistics and transformational grammar were exploited to improve the quality of translations. During this period operational systems were installed. The United States Air Force used a system produced by IBM and Washington University in St. Louis, while the Atomic Energy Commission and Euratom, in Italy, used a system developed at Georgetown University. While the quality of the output was poor it met many of the customers' needs, particularly in terms of speed. At the end of the 1950s, Yehoshua Bar-Hillel was asked by the US government to look into machine translation, to assess the possibility of fully automatic high-quality translation by machines. Bar-Hillel described the problem of semantic ambiguity or double-meaning, as illustrated in the following sentence: Little John was looking for his toy box. Finally he found it. The box was in the pen. The word pen may have two meanings: the first meaning, something used to write in ink with; the second meaning, a container of some kind. To a human, the meaning is obvious, but Bar-Hillel claimed that without a "universal encyclopedia" a machine would never be able to deal with this problem. At the time, this type of semantic ambiguity could only be solved by writing source texts for machine translation in a controlled language that uses a vocabulary in which each word has exactly one meaning. == The 1960s, the ALPAC report and the seventies == Research in the 1960s in both the Soviet Union and the United States concentrated mainly on the Russian–English language pair. The objects of translation were chiefly scientific and technical documents, such as articles from scientific journals. The rough translations produced were sufficient to get a basic understanding of the articles. If an article discussed a subject deemed to be confidential, it was sent to a human translator for a complete translation; if not, it was discarded. A great blow came to machine-translation research in 1966 with the publication of the ALPAC report. The report was commissioned by the US government and delivered by ALPAC, the Automatic Language Processing Advisory Committee, a group of seven scientists convened by the US government in 1964. The US government was concerned that there was a lack of progress being made despite significant expenditure. The report concluded that machine translation was more expensive, less accurate and slower than human translation, and that despite the expenditures, machine translation was not likely to reach the quality of a human translator in the near future. The report recommended, however, that tools be developed to aid translators – automatic dictionaries, for example – and that some research in computational linguistics should continue to be supported. The publication of the report had a profound impact on research into machine translation in the United States, and to a lesser extent the Soviet Union and United Kingdom. Research, at least in the US, was almost completely abandoned for over a decade. In Canada, France and Germany, however, research continued. In the US the main exceptions were the founders of SYSTRAN (Peter Toma) and Logos (Bernard Scott), who established their companies in 1968 and 1970 respectively and served the US Department of Defense. In 1970, the SYSTRAN system was installed for the United States Air Force, and subsequently by the Commission of the European Communities in 1976. The METEO System, developed at the Université de Montréal, was installed in Canada in 1977 to translate weather forecasts from English to French, and was translating close to 80,000 words per day or 30 million words per year until it was replaced by a competitor's system on 30 September 2001. While research in the 1960s concentrated on limited language pairs and input, demand in the 1970s was for low-cost systems that could translate a range of technical and commercial documents. This demand was spurred by the increase of globalisation and the demand for translation in Canada, Europe, and Japan. == The 1980s and early 1990s == By the 1980s, both the diversity and the number of installed systems for machine translation had increased. A number of systems relying on mainframe technology were in use, such as SYSTRAN, Logos, Ariane-G5, and Metal. As a result of the improved availability of microcomputers, there was a market for lower-end machine translation systems. Many companies took advantage of this in Europe, Japan, and the USA. Systems were also brought onto the market in China, Eastern Europe, Korea, and the Soviet Union. During the 1980s there was a lot of activity in MT in Japan especially. With the fifth-generation co

  Read more →
• 

  DeepSeek (chatbot)

  DeepSeek is a generative artificial intelligence chatbot developed by the Chinese company DeepSeek. Released on 20 January 2025, DeepSeek-R1 surpassed ChatGPT as the most downloaded freeware app on the iOS App Store in the United States by 27 January. DeepSeek's success against larger and more established rivals has been described as "upending AI" and initiating "a global AI space race". DeepSeek's compliance with Chinese government censorship policies and its data collection practices have also raised concerns over privacy and information control in the model, prompting regulatory scrutiny in multiple countries. However, it has also been praised for its open weights and infrastructure code, energy efficiency and contributions to open-source artificial intelligence. == History == On 10 January 2025, DeepSeek released the chatbot, based on the DeepSeek-R1 model, for iOS and Android. By 27 January, DeepSeek-R1 surpassed ChatGPT as the most-downloaded freeware app on the iOS App Store in the United States, which resulted in an 18% drop in Nvidia's share price. And after a "large-scale" cyberattack on the same day disrupted the proper functioning of its servers, DeepSeek had limited its new user registration to phone numbers from mainland China, email addresses, or Google account logins. On 3 April 2025, in collaboration with researchers at Tsinghua University, DeepSeek published a paper unveiling a new model that combines the techniques generative reward modeling (GRM) and self-principled critique tuning (SPCT). The resulting model is referred to as DeepSeek-GRM. The goal of using these techniques is to foster more effective inference-time scaling within their LLM and chatbot services. Notably, DeepSeek has said that these new models will be released and made open source. On 30 April 2025, Deepseek released its math-focused Artificial Intelligence Model named "DeepSeek-Prover-V2-671B". This model is useful for formal theorem proving and mathematical reasoning. On 24 April 2026, DeepSeek released DeepSeek V4 and V4-Pro. == Usage == DeepSeek can answer questions, solve logic problems, and write computer programs on par with other chatbots, according to benchmark tests used by American AI companies. Users can access the chatbot for free through the official DeepSeek website or mobile application, without limitation on the number of queries. DeepSeek only supports user-signup via a global email service, e.g. Gmail, Google or Yahoo. DeepSeek also offers access to the R1 and V3 models that power the chatbot via an API with a usage-based pricing model. This modality is primarily targeted towards developers and businesses. As of February 2025, API usage is priced at approximately $0.28 per million input tokens and $0.42 per million output tokens, making it less expensive than some competing services. Its web version is completely free, with 500 messages per hour cap limit to prevent bots from spamming. == Operation == DeepSeek-V3 uses significantly fewer resources compared to its peers. For example, whereas the world's leading AI companies train their chatbots with supercomputers using as many as 16,000 graphics processing units (GPUs), DeepSeek claims to have needed only about 2,000 GPUs—namely, the H800 series chips from Nvidia. It was trained in around 55 days at a cost of US$5.58 million, which is roughly one-tenth of what tech giant Meta spent building its latest AI technology. == Reactions == DeepSeek's success against larger and more established rivals has been described as "upending AI", constituting "the first shot at what is emerging as a global AI space race", and ushering in "a new era of AI brinkmanship". === Challenge to US AI dominance === DeepSeek's competitive performance at relatively minimal cost has been recognized as potentially challenging the global dominance of American AI models. Various publications and news media, such as The Hill and The Guardian, have described the release of the R1 chatbot as a "Sputnik moment" for American AI, echoing Marc Andreessen's view. OpenAI wrote a letter to the Office of Science and Technology Policy (OSTP), in March 2025, citing issues concerning a possibility that Deepseek could manipulate responses to cause harm. === Chinese perspective === DeepSeek's founder Liang Wenfeng has been compared to OpenAI CEO Sam Altman, with CNN calling him the Sam Altman of China and an evangelist for AI. Chinese state media widely praised DeepSeek as a national asset. On 20 January 2025, Chinese Premier Li Qiang invited Wenfeng to his symposium with experts and asked him to provide opinions and suggestions on a draft for comments of the annual 2024 government work report. On 20 February 2025, Wenfeng met with General Secretary of the Chinese Communist Party Xi Jinping, who encouraged party and state leaders to experiment with DeepSeek. Government officials responded to Xi's approval of the chatbot by reportedly using it to draft legal judgements, propose medical treatment plans, and analyze surveillance videos to search for missing persons. === Performance and success === Leading figures in the American AI sector had mixed reactions to DeepSeek's performance and success. Microsoft CEO Satya Nadella and OpenAI CEO Altman—whose companies are involved in the United States government-backed "Stargate Project" to develop American AI infrastructure—both called DeepSeek "super impressive". Various companies including Amazon Web Services, Toyota, and Stripe are seeking to use the model in their program. When American President Donald Trump announced The Stargate Project, he referred to DeepSeek as a wake-up call and a positive development. Other leaders in the AI field, however—including Scale AI CEO Alexandr Wang, Anthropic cofounder and CEO Dario Amodei, and Elon Musk—have expressed skepticism of the app's performance or of the sustainability of its success. Wang in particularly referred to DeepSeek-V3 as "earth-shattering" and DeepSeek-R1 as "top performing, or roughly on par with the best American models", but speculated that China may possess more AI-powering Nvidia H100 GPUs than thought. === Stock market implications === DeepSeek's optimization of limited resources has highlighted potential limits of United States sanctions on China's AI development, including export restrictions on advanced AI chips to China. The success of the company's AI models consequently "sparked market turmoil" and caused shares in major global technology companies to plunge on 27 January 2025: Nvidia's stock fell by as much as 17–18%, as did the stock of rival Broadcom. Other tech firms also sank, including Microsoft (down 2.5%), Google's owner Alphabet (down over 4%), and Dutch chip equipment maker ASML (down over 7%). A global sell-off of technology stocks on Nasdaq, prompted by the release of the R1 model, led to record losses of about $593 billion in the market capitalizations of AI and computer hardware companies; and by the next day a total of $1 trillion of value was wiped from American stocks. == Concerns == === Distillation === DeepSeek has been reported to sometimes claim that it is ChatGPT. OpenAI said that DeepSeek may have "inappropriately" used outputs from its model as training data in a process called distillation. However, there is currently no method to prove this conclusively. === Censorship === DeepSeek's compliance with Chinese government censorship policies and its data collection practices have raised concerns over information control in the model, prompting regulatory scrutiny in multiple countries. Reports indicate that it applies content moderation in accordance with the government's "public opinion guidance" regulations, limiting responses on topics such as the Tiananmen Square massacre and Taiwan's political status. DeepSeek models that have been uncensored also display a bias towards Chinese government viewpoints on controversial topics such as Xi Jinping's human rights record and Taiwan's political status. However, users who have downloaded the models and hosted them on their own devices and servers have reported successfully removing this censorship. Some sources have observed that the official application programming interface (API) version of R1, which runs from servers located in mainland China, uses censorship mechanisms for topics considered politically sensitive for the government of China. For example, the model may initially generate answers to questions about the 1989 Tiananmen Square massacre, persecution of Uyghurs, comparisons between Xi Jinping and Winnie the Pooh, and human rights in China, but a censorship mechanism deletes the uncensored response afterwards and replaces it with a message such as:"Sorry, that's beyond my current scope. Let's talk about something else." The post hoc censorship mechanisms and restrictions added on top of the model's output can be removed in the open-source version of the R1 model. If the "core Socialist values" defined by the Chinese Internet regul

  Read more →
• 

  Sentence embedding

  In natural language processing, a sentence embedding is a representation of a sentence as a vector of numbers which encodes meaningful semantic information. State of the art embeddings are based on the learned hidden layer representation of dedicated sentence transformer models. BERT pioneered an approach involving the use of a dedicated [CLS] token prepended to the beginning of each sentence inputted into the model; the final hidden state vector of this token encodes information about the sentence and can be fine-tuned for use in sentence classification tasks. In practice however, BERT's sentence embedding with the [CLS] token achieves poor performance, often worse than simply averaging non-contextual word embeddings. SBERT later achieved superior sentence embedding performance by fine tuning BERT's [CLS] token embeddings through the usage of a siamese neural network architecture on the SNLI dataset. Other approaches are loosely based on the idea of distributional semantics applied to sentences. Skip-Thought trains an encoder-decoder structure for the task of neighboring sentences predictions; this has been shown to achieve worse performance than approaches such as InferSent or SBERT. An alternative direction is to aggregate word embeddings, such as those returned by Word2vec, into sentence embeddings. The most straightforward approach is to simply compute the average of word vectors, known as continuous bag-of-words (CBOW). However, more elaborate solutions based on word vector quantization have also been proposed. One such approach is the vector of locally aggregated word embeddings (VLAWE), which demonstrated performance improvements in downstream text classification tasks. == Applications == In recent years, sentence embedding has seen a growing level of interest due to its applications in natural language queryable knowledge bases through the usage of vector indexing for semantic search. LangChain for instance utilizes sentence transformers for purposes of indexing documents. In particular, an indexing is generated by generating embeddings for chunks of documents and storing (document chunk, embedding) tuples. Then given a query in natural language, the embedding for the query can be generated. A top k similarity search algorithm is then used between the query embedding and the document chunk embeddings to retrieve the most relevant document chunks as context information for question answering tasks. This approach is also known formally as retrieval-augmented generation. Though not as predominant as BERTScore, sentence embeddings are commonly used for sentence similarity evaluation which sees common use for the task of optimizing a Large language model's generation parameters is often performed via comparing candidate sentences against reference sentences. By using the cosine-similarity of the sentence embeddings of candidate and reference sentences as the evaluation function, a grid-search algorithm can be utilized to automate hyperparameter optimization. == Evaluation == A way of testing sentence encodings is to apply them on Sentences Involving Compositional Knowledge (SICK) corpus for both entailment (SICK-E) and relatedness (SICK-R). In the best results are obtained using a BiLSTM network trained on the Stanford Natural Language Inference (SNLI) Corpus. The Pearson correlation coefficient for SICK-R is 0.885 and the result for SICK-E is 86.3. A slight improvement over previous scores is presented in: SICK-R: 0.888 and SICK-E: 87.8 using a concatenation of bidirectional Gated recurrent unit.

  Read more →
• 

  Cuboid (computer vision)

  In computer vision, the term cuboid is used to describe a small spatiotemporal volume extracted for purposes of behavior recognition. The cuboid is regarded as a basic geometric primitive type and is used to depict three-dimensional objects within a three dimensional representation of a flat, two dimensional image. == Production == Cuboids can be produced from both two-dimensional and three-dimensional images. One method used to produce cuboids utilizes scene understanding (SUN) primitive databases, which are collections of pictures that already contain cuboids. By sorting through SUN primitive databases with machine learning tools, computers observe the conditions in which cuboids are produced in images from SUN primitive databases and can learn to produce cuboids from other images. RGB-D images, which are RGB images that also record the depth of each pixel, are occasionally used to produce cuboids because computers no longer need to determine the depth of an object, as they typically do because depth is already recorded. Cuboid production is sensitive to changes in color and illumination, blockage, and background clutter. This means that it is difficult for computers to produce cuboids of objects that are multicolored, irregularly illuminated, or partially covered, or if there are many objects in the background. This is partially due to the fact that algorithms for producing cuboids are still relatively simple. == Usage == Cuboids are created for point cloud-based three-dimensional maps and can be utilized in various situations such as augmented reality, the automated control of cars, drones, and robots, and object detection. Cuboids allow for software to identify a scene through geometric descriptions in an “object-agnostic” fashion. Interest points, locations within images that are identified by a computer as essential to identifying the image, created from two-dimensional images can be used with cuboids for image matching, identifying a room or scene, and instance recognition. Interest points created from three dimensional images can be used with cuboids to recognize activities. This is possible because interest points aid software to focus on only the most important aspects of the images. RGB-D images and SLAM systems are used together in RGB-D SLAM systems, which are employed by Computer-aided design systems to generate point cloud-based three-dimensional maps. Most industrial multi-axis machining tools use computer-aided manufacturing and subsequently work in cuboid work spaces.

  Read more →
• 

  Kleene star

  In formal language theory, the Kleene star (or Kleene operator or Kleene closure) refers to two related unary operations, that can be applied either to an alphabet of symbols or to a formal language, a set of strings (finite sequences of symbols). The Kleene star operator on an alphabet V generates the set V of all finite-length strings over V, that is, finite sequences whose elements belong to V; in mathematics, it is more commonly known as the free monoid construction. The Kleene star operator on a language L generates another language L, the set of all strings that can be obtained as a concatenation of zero or more members of L. In both cases, repetitions are allowed. The Kleene star operators are named after American mathematician Stephen Cole Kleene, who first introduced and widely used it to characterize automata for regular expressions. == Of an alphabet == Given an alphabet V {\displaystyle V} , define V 0 = { ε } {\displaystyle V^{0}=\{\varepsilon \}} (the set consists only of the empty string), V 1 = V , {\displaystyle V^{1}=V,} and define recursively the set V i + 1 = { w v : w ∈ V i and v ∈ V } {\displaystyle V^{i+1}=\{wv:w\in V^{i}{\text{ and }}v\in V\}} for each i > 0 , {\displaystyle i>0,} where w v {\displaystyle wv} denotes the string obtained by appending the single character v {\displaystyle v} to the end of w {\displaystyle w} . Here, V i {\displaystyle V^{i}} can be understood to be the set of all strings of length exactly i {\displaystyle i} , with characters from V {\displaystyle V} . The definition of Kleene star on V {\displaystyle V} is V ∗ = ⋃ i ≥ 0 V i = V 0 ∪ V 1 ∪ V 2 ∪ V 3 ∪ V 4 ∪ ⋯ . {\displaystyle V^{}=\bigcup _{i\geq 0}V^{i}=V^{0}\cup V^{1}\cup V^{2}\cup V^{3}\cup V^{4}\cup \cdots .} == Of a language == Given a language L {\displaystyle L} (any finite or infinite set of strings), define L 0 = { ε } {\displaystyle L^{0}=\{\varepsilon \}} (the language consisting only of the empty string), L 1 = L , {\displaystyle L^{1}=L,} and define recursively the set L i + 1 = { w v : w ∈ L i and v ∈ L } {\displaystyle L^{i+1}=\{wv:w\in L^{i}{\text{ and }}v\in L\}} for each i > 0 , {\displaystyle i>0,} where w v {\displaystyle wv} denotes the string obtained by concatenating w {\displaystyle w} and v {\displaystyle v} . Here, L i {\displaystyle L^{i}} can be understood to be the set of all strings that can be obtained by concatenating exactly i {\displaystyle i} strings from L {\displaystyle L} , allowing repetitions. The definition of Kleene star on L {\displaystyle L} is L ∗ = ⋃ i ≥ 0 L i = L 0 ∪ L 1 ∪ L 2 ∪ L 3 ∪ L 4 ∪ ⋯ . {\displaystyle L^{}=\bigcup _{i\geq 0}L^{i}=L^{0}\cup L^{1}\cup L^{2}\cup L^{3}\cup L^{4}\cup \cdots .} == Kleene plus == In some formal language studies, (e.g. AFL theory) a variation on the Kleene star operation called the Kleene plus is used. The Kleene plus omits the V 0 {\displaystyle V^{0}} or L 0 {\displaystyle L^{0}} term in the above unions. In other words, the Kleene plus on V {\displaystyle V} is V + = ⋃ i ≥ 1 V i = V 1 ∪ V 2 ∪ V 3 ∪ ⋯ , {\displaystyle V^{+}=\bigcup _{i\geq 1}V^{i}=V^{1}\cup V^{2}\cup V^{3}\cup \cdots ,} or V + = V ∗ V . {\displaystyle V^{+}=V^{}V.} == Examples == Example of Kleene star applied to a set of strings: {"ab","c"} = { ε, "ab", "c", "abab", "abc", "cab", "cc", "ababab", "ababc", "abcab", "abcc", "cabab", "cabc", "ccab", "ccc", ...}. Example of Kleene star applied to a set of strings without the prefix property: {"a","ab","b"} = { ε, "a", "ab", "b", "aa", "aab", "aba", "abab", "abb", "ba", "bab", "bb", ...};In this example, the string "aab" can be obtained in two different ways. The Sardinas-Patterson algorithm can be used to check for a given V whether any member of V can be obtained in more than one way. Example of Kleene and Kleene plus applied to a set of characters (following the C programming language convention where a character is denoted by single quotes and a string is denoted by double quotes): {'a', 'b', 'c'} = { ε, "a", "b", "c", "aa", "ab", "ac", "ba", "bb", "bc", "ca", "cb", "cc", "aaa", "aab", ...}. {'a', 'b', 'c'}+ = { "a", "b", "c", "aa", "ab", "ac", "ba", "bb", "bc", "ca", "cb", "cc", "aaa", "aab", ...}. == Properties == If V {\displaystyle V} is any finite or countably infinite set of characters, then V ∗ {\displaystyle V^{}} is a countably infinite set. As a result, each formal language over a finite or countably infinite alphabet Σ {\displaystyle \Sigma } is countable, since it is a subset of the countably infinite set Σ ∗ {\displaystyle \Sigma ^{}} . ( L ∗ ) ∗ = L ∗ {\displaystyle (L^{})^{}=L^{}} , which means that the Kleene star operator is an idempotent unary operator, as ( L ∗ ) i = L ∗ {\displaystyle (L^{})^{i}=L^{}} for every i ≥ 1 {\displaystyle i\geq 1} . V ∗ = { ε } {\displaystyle V^{}=\{\varepsilon \}} , if V {\displaystyle V} is the empty set ∅. For the version of the Kleene star operator on languages, L ∗ = { ε } {\displaystyle L^{}=\{\varepsilon \}} when L {\displaystyle L} is either the empty set ∅ or the singleton set { ε } {\displaystyle \{\varepsilon \}} . == Generalization == Strings form a monoid with concatenation as the binary operation and ε the identity element. In addition to strings, the Kleene star is defined for any monoid. More precisely, let (M, ⋅) be a monoid, and S ⊆ M. Then S is the smallest submonoid of M containing S; that is, S contains the neutral element of M, the set S, and is such that if x,y ∈ S, then x⋅y ∈ S. Furthermore, the Kleene star is generalized by including the -operation (and the union) in the algebraic structure itself by the notion of complete star semiring.

  Read more →
• 

  Dr. Sbaitso

  Dr. Sbaitso ( SPAYT-soh) is an artificial intelligence speech synthesis program released late in 1991 by Creative Labs in Singapore for MS-DOS-based personal computers. The name is an acronym for "SoundBlaster Acting Intelligent Text-to-Speech Operator." == History == Dr. Sbaitso was distributed with various sound cards manufactured by Creative Technology in the early 1990s. The text-to-speech engine used is a version of Monologue, which was developed by First Byte Software. Monologue is a later release of First Byte's "SmoothTalker" software from 1984. The program "conversed" with the user as if it were a psychologist, though most of its responses were along the lines of "WHY DO YOU FEEL THAT WAY?" rather than any sort of complicated interaction. When confronted with a phrase it could not understand, it would often reply with something such as "THAT'S NOT MY PROBLEM." Dr. Sbaitso repeated text out loud that was typed after the word "SAY." Repeated swearing or abusive behavior on the part of the user caused Dr. Sbaitso to "break down" in a "PARITY ERROR" before resetting itself. The same would happen, if the user types "SAY PARITY." The program introduced itself with the following lines: HELLO [UserName], MY NAME IS DOCTOR SBAITSO. I AM HERE TO HELP YOU. SAY WHATEVER IS IN YOUR MIND FREELY, OUR CONVERSATION WILL BE KEPT IN STRICT CONFIDENCE. MEMORY CONTENTS WILL BE WIPED OFF AFTER YOU LEAVE, SO, TELL ME ABOUT YOUR PROBLEMS. The program was designed to showcase the digitized voices the cards were able to produce, though the quality was far from lifelike. Additionally, there was a version of this program for Microsoft Windows through the use of a program called Prody Parrot; this version of the software featured a more detailed graphical user interface. The text-to-speech was also used as the voice of 1st Prize from the Baldi's Basics series, albeit slowed down. == Commands == If the user submits "HELP", a list of commands will appear. If the user then submits "M", more commands will appear. There are three pages of commands in total, with guidance on how to use each of the features.

  Read more →
• 

  Tesla Dojo

  Tesla Dojo is a series of supercomputers designed and built by Tesla for computer vision video processing and recognition. It was used for training Tesla's machine learning models to improve its Full Self-Driving (FSD) advanced driver-assistance system. It went into production in July 2023. Dojo's goal was to efficiently process millions of terabytes of video data captured from real-life driving situations from Tesla's 4+ million cars. This goal led to a considerably different architecture than conventional supercomputer designs. In August 2025, Bloomberg News reported that the Dojo project had been disbanded, though it was restarted in January 2026. == History == Tesla operates several massively parallel computing clusters for developing its Autopilot advanced driver assistance system. Its primary unnamed cluster using 5,760 Nvidia A100 graphics processing units (GPUs) was touted by Andrej Karpathy in 2021 at the fourth International Joint Conference on Computer Vision and Pattern Recognition (CCVPR 2021) to be "roughly the number five supercomputer in the world" at approximately 81.6 petaflops, based on scaling the performance of the Nvidia Selene supercomputer, which uses similar components. However, the performance of the primary Tesla GPU cluster has been disputed, as it was not clear if this was measured using single-precision or double-precision floating point numbers (FP32 or FP64). Tesla also operates a second 4,032 GPU cluster for training and a third 1,752 GPU cluster for automatic labeling of objects. The primary unnamed Tesla GPU cluster has been used for processing one million video clips, each ten seconds long, taken from Tesla Autopilot cameras operating in Tesla cars in the real world, running at 36 frames per second. Collectively, these video clips contained six billion object labels, with depth and velocity data; the total size of the data set was 1.5 petabytes. This data set was used for training a neural network intended to help Autopilot computers in Tesla cars understand roads. By August 2022, Tesla had upgraded the primary GPU cluster to 7,360 GPUs. Dojo was first mentioned by Elon Musk in April 2019 during Tesla's "Autonomy Investor Day". In August 2020, Musk stated it was "about a year away" due to power and thermal issues. Dojo was officially announced at Tesla's Artificial Intelligence (AI) Day on August 19, 2021. Tesla revealed details of the D1 chip and its plans for "Project Dojo", a datacenter that would house 3,000 D1 chips; the first "Training Tile" had been completed and delivered the week before. In October 2021, Tesla released a "Dojo Technology" whitepaper describing the Configurable Float8 (CFloat8) and Configurable Float16 (CFloat16) floating point formats and arithmetic operations as an extension of Institute of Electrical and Electronics Engineers (IEEE) standard 754. At the follow-up AI Day in September 2022, Tesla announced it had built several System Trays and one Cabinet. During a test, the company stated that Project Dojo drew 2.3 megawatts (MW) of power before tripping a local San Jose, California power substation. At the time, Tesla was assembling one Training Tile per day. In August 2023, Tesla powered on Dojo for production use as well as a new training cluster configured with 10,000 Nvidia H100 GPUs. In January 2024, Musk described Dojo as "a long shot worth taking because the payoff is potentially very high. But it's not something that is a high probability." In June 2024, Musk explained that ongoing construction work at Gigafactory Texas is for a computing cluster claiming that it is planned to comprise an even mix of "Tesla AI" and Nvidia/other hardware with a total thermal design power of at first 130 MW and eventually exceeding 500 MW. In August 2025, Bloomberg News reported that the Dojo project was disbanded, though Musk announced it would be restarted in January 2026 with a new chip iteration. == Technical architecture == The fundamental unit of the Dojo supercomputer is the D1 chip, designed by a team at Tesla led by ex-AMD CPU designer Ganesh Venkataramanan, including Emil Talpes, Debjit Das Sarma, Douglas Williams, Bill Chang, and Rajiv Kurian. The D1 chip is manufactured by the Taiwan Semiconductor Manufacturing Company (TSMC) using 7 nanometer (nm) semiconductor nodes, has 50 billion transistors and a large die size of 645 mm2 (1.0 square inch). Updating at Artificial Intelligence (AI) Day in 2022, Tesla announced that Dojo would scale by deploying multiple ExaPODs, in which there would be: 10 Cabinets per ExaPOD (1,062,000 cores, 3,000 D1 chips) 2 System Trays per Cabinet (106,200 cores, 300 D1 chips) 6 Training Tiles per System Tray (53,100 cores, along with host interface hardware) 25 D1 chips per Training Tile (8,850 cores) 354 computing cores per D1 chip According to Venkataramanan, Tesla's senior director of Autopilot hardware, Dojo will have more than an exaflop (a million teraflops) of computing power. For comparison, according to Nvidia, in August 2021, the (pre-Dojo) Tesla AI-training center used 720 nodes, each with eight Nvidia A100 Tensor Core GPUs for 5,760 GPUs in total, providing up to 1.8 exaflops of performance. === D1 chip === Each node (computing core) of the D1 processing chip is a general purpose 64-bit CPU with a superscalar core. It supports internal instruction-level parallelism, and includes simultaneous multithreading (SMT). It doesn't support virtual memory and uses limited memory protection mechanisms. Dojo software/applications manage chip resources. The D1 instruction set supports both 64-bit scalar and 64-byte single instruction, multiple data (SIMD) vector instructions. The integer unit mixes reduced instruction set computer (RISC-V) and custom instructions, supporting 8, 16, 32, or 64 bit integers. The custom vector math unit is optimized for machine learning kernels and supports multiple data formats, with a mix of precisions and numerical ranges, many of which are compiler composable. Up to 16 vector formats can be used simultaneously. ==== Node ==== Each D1 node uses a 32-byte fetch window holding up to eight instructions. These instructions are fed to an eight-wide decoder which supports two threads per cycle, followed by a four-wide, four-way SMT scalar scheduler that has two integer units, two address units, and one register file per thread. Vector instructions are passed further down the pipeline to a dedicated vector scheduler with two-way SMT, which feeds either a 64-byte SIMD unit or four 8×8×4 matrix multiplication units. The network on-chip (NOC) router links cores into a two-dimensional mesh network. It can send one packet in and one packet out in all four directions to/from each neighbor node, along with one 64-byte read and one 64-byte write to local SRAM per clock cycle. Hardware native operations transfer data, semaphores and barrier constraints across memories and CPUs. System-wide double data rate 4 (DDR4) synchronous dynamic random-access memory (SDRAM) memory works like bulk storage. ==== Memory ==== Each core has a 1.25 megabytes (MB) of SRAM main memory. Load and store speeds reach 400 gigabytes (GB) per second and 270 GB/sec, respectively. The chip has explicit core-to-core data transfer instructions. Each SRAM has a unique list parser that feeds a pair of decoders and a gather engine that feeds the vector register file, which together can directly transfer information across nodes. ==== Die ==== Twelve nodes (cores) are grouped into a local block. Nodes are arranged in an 18×20 array on a single die, of which 354 cores are available for applications. The die runs at 2 gigahertz (GHz) and totals 440 MB of SRAM (360 cores × 1.25 MB/core). It reaches 376 teraflops using 16-bit brain floating point (BF16) numbers or using configurable 8-bit floating point (CFloat8) numbers, which is a Tesla proposal, and 22 teraflops at FP32. Each die comprises 576 bi-directional serializer/deserializer (SerDes) channels along the perimeter to link to other dies, and moves 8 TB/sec across all four die edges. Each D1 chip has a thermal design power of approximately 400 watts. === Training Tile === The water-cooled Training Tile packages 25 D1 chips into a 5×5 array. Each tile supports 36 TB/sec of aggregate bandwidth via 40 input/output (I/O) chips - half the bandwidth of the chip mesh network. Each tile supports 10 TB/sec of on-tile bandwidth. Each tile has 11 GB of SRAM memory (25 D1 chips × 360 cores/D1 × 1.25 MB/core). Each tile achieves 9 petaflops at BF16/CFloat8 precision (25 D1 chips × 376 TFLOP/D1). Each tile consumes 15 kilowatts; 288 amperes at 52 volts. === System Tray === Six tiles are aggregated into a System Tray, which is integrated with a host interface. Each host interface includes 512 x86 cores, providing a Linux-based user environment. Previously, the Dojo System Tray was known as the Training Matrix, which includes six Training Tiles, 20 Dojo Interface Processor cards across four host servers, and Ethernet-l

  Read more →
• 

  Convolutional neural network

  A convolutional neural network (CNN) is a type of feedforward neural network that learns features via filter (or kernel) optimization. This type of deep learning network has been applied to process and make predictions from many different types of data including text, images and audio. CNNs are the de-facto standard in deep learning-based approaches to computer vision and image processing, and have only recently been replaced—in some cases—by newer architectures such as the transformer. Vanishing gradients and exploding gradients, seen during backpropagation in earlier neural networks, are prevented by the regularization that comes from using shared weights over fewer connections. For example, for each neuron in the fully-connected layer, 10,000 weights would be required for processing an image sized 100 × 100 pixels. However, applying cascaded convolution (or cross-correlation) kernels, only 25 weights for each convolutional layer are required to process 5x5-sized tiles. Higher-layer features are extracted from wider context windows, compared to lower-layer features. Some applications of CNNs include: image and video recognition, recommender systems, image classification, image segmentation, medical image analysis, natural language processing, brain–computer interfaces, and financial time series. CNNs are also known as shift invariant or space invariant artificial neural networks, based on the shared-weight architecture of the convolution kernels or filters that slide along input features and provide translation-equivariant responses known as feature maps. Counter-intuitively, most convolutional neural networks are not invariant to translation, due to the downsampling operation they apply to the input. Feedforward neural networks are usually fully connected networks, that is, each neuron in one layer is connected to all neurons in the next layer. The "full connectivity" of these networks makes them prone to overfitting data. Typical ways of regularization, or preventing overfitting, include: penalizing parameters during training (such as weight decay) or trimming connectivity (skipped connections, dropout, etc.) Robust datasets also increase the probability that CNNs will learn the generalized principles that characterize a given dataset rather than the biases of a poorly-populated set. Convolutional networks were inspired by biological processes in that the connectivity pattern between neurons resembles the organization of the animal visual cortex. Individual cortical neurons respond to stimuli only in a restricted region of the visual field known as the receptive field. The receptive fields of different neurons partially overlap such that they cover the entire visual field. CNNs use relatively little pre-processing compared to other image classification algorithms. This means that the network learns to optimize the filters (or kernels) through automated learning, whereas in traditional algorithms these filters are hand-engineered. This simplifies and automates the process, enhancing efficiency and scalability overcoming human-intervention bottlenecks. == Architecture == A convolutional neural network consists of an input layer, hidden layers and an output layer. In a convolutional neural network, the hidden layers include one or more layers that perform convolutions. Typically this includes a layer that performs a dot product of the convolution kernel with the layer's input matrix. This product is usually the Frobenius inner product, and its activation function is commonly ReLU. As the convolution kernel slides along the input matrix for the layer, the convolution operation generates a feature map, which in turn contributes to the input of the next layer. This is followed by other layers such as pooling layers, fully connected layers, and normalization layers. Here it should be noted how close a convolutional neural network is to a matched filter. === Convolutional layers === In a CNN, the input is a tensor with shape: (number of inputs) × (input height) × (input width) × (input channels) After passing through a convolutional layer, the image becomes abstracted to a feature map, also called an activation map, with shape: (number of inputs) × (feature map height) × (feature map width) × (feature map channels). Convolutional layers convolve the input and pass its result to the next layer. This is similar to the response of a neuron in the visual cortex to a specific stimulus. Each convolutional neuron processes data only for its receptive field. Although fully connected feedforward neural networks can be used to learn features and classify data, this architecture is generally impractical for larger inputs (e.g., high-resolution images), which would require massive numbers of neurons because each pixel is a relevant input feature. A fully connected layer for an image of size 100 × 100 has 10,000 weights for each neuron in the second layer. Convolution reduces the number of free parameters, allowing the network to be deeper. For example, using a 5 × 5 tiling region, each with the same shared weights, requires only 25 neurons. Using shared weights means there are many fewer parameters, which helps avoid the vanishing gradients and exploding gradients problems seen during backpropagation in earlier neural networks. To speed processing, standard convolutional layers can be replaced by depthwise separable convolutional layers, which are based on a depthwise convolution followed by a pointwise convolution. The depthwise convolution is a spatial convolution applied independently over each channel of the input tensor, while the pointwise convolution is a standard convolution restricted to the use of 1 × 1 {\displaystyle 1\times 1} kernels. === Pooling layers === Convolutional networks may include local and/or global pooling layers along with traditional convolutional layers. Pooling layers reduce the dimensions of data by combining the outputs of neuron clusters at one layer into a single neuron in the next layer. Local pooling combines small clusters, tiling sizes such as 2 × 2 are commonly used. Global pooling acts on all the neurons of the feature map. There are two common types of pooling in popular use: max and average. Max pooling uses the maximum value of each local cluster of neurons in the feature map, while average pooling takes the average value. === Fully connected layers === Fully connected layers connect every neuron in one layer to every neuron in another layer. It is the same as a traditional multilayer perceptron neural network (MLP). Each neuron in the fully connected layer receives input from all the neurons in the previous layer. These inputs are weighted and summed with the corresponding biases, and then passed through an activation function to perform a nonlinear transformation, generating the output. The flattened matrix goes through a fully connected layer to classify the images. === Receptive field === In neural networks, each neuron receives input from some number of locations in the previous layer. In a convolutional layer, each neuron receives input from only a restricted area of the previous layer called the neuron's receptive field. Typically the area is a square (e.g. 5 by 5 neurons). Whereas, in a fully connected layer, the receptive field is the entire previous layer. Thus, in each convolutional layer, each neuron takes input from a larger area in the input than previous layers. This is due to applying the convolution over and over, which takes the value of a pixel into account, as well as its surrounding pixels. When using dilated layers, the number of pixels in the receptive field remains constant, but the field is more sparsely populated as its dimensions grow when combining the effect of several layers. To manipulate the receptive field size as desired, there are some alternatives to the standard convolutional layer. For example, atrous or dilated convolution expands the receptive field size without increasing the number of parameters by interleaving visible and blind regions. Moreover, a single dilated convolutional layer can comprise filters with multiple dilation ratios, thus having a variable receptive field size. === Weights === Each neuron in a neural network computes an output value by applying a specific function to the input values received from the receptive field in the previous layer. The function that is applied to the input values is determined by a vector of weights and a bias (typically real numbers). Learning consists of iteratively adjusting these biases and weights. The vectors of weights and biases are called filters and represent particular features of the input (e.g., a particular shape). A distinguishing feature of CNNs is that many neurons can share the same filter. This reduces the memory footprint because a single bias and a single vector of weights are used across all receptive fields that share that filter, as opposed to each receptive field having its own bias and vector

  Read more →
• 

  Teknomo–Fernandez algorithm

  The Teknomo–Fernandez algorithm (TF algorithm), is an efficient algorithm for generating the background image of a given video sequence. By assuming that the background image is shown in the majority of the video, the algorithm is able to generate a good background image of a video in O ( R ) {\displaystyle O(R)} -time using only a small number of binary operations and Boolean bit operations, which require a small amount of memory and has built-in operators found in many programming languages such as C, C++, and Java. == History == People tracking from videos usually involves some form of background subtraction to segment foreground from background. Once foreground images are extracted, then desired algorithms (such as those for motion tracking, object tracking, and facial recognition) may be executed using these images. However, background subtraction requires that the background image is already available and unfortunately, this is not always the case. Traditionally, the background image is searched for manually or automatically from the video images when there are no objects. More recently, automatic background generation through object detection, medial filtering, medoid filtering, approximated median filtering, linear predictive filter, non-parametric model, Kalman filter, and adaptive smoothening have been suggested; however, most of these methods have high computational complexity and are resource-intensive. The Teknomo–Fernandez algorithm is also an automatic background generation algorithm. Its advantage, however, is its computational speed of only O ( R ) {\displaystyle O(R)} -time, depending on the resolution R {\displaystyle R} of an image and its accuracy gained within a manageable number of frames. Only at least three frames from a video is needed to produce the background image assuming that for every pixel position, the background occurs in the majority of the videos. Furthermore, it can be performed for both grayscale and colored videos. == Assumptions == The camera is stationary. The light of the environment changes only slowly relative to the motions of the people in the scene. The number of people does not occupy the scene for most of the time at the same place. Generally, however, the algorithm will certainly work whenever the following single important assumption holds: For each pixel position, the majority of the pixel values in the entire video contain the pixel value of the actual background image (at that position).As long as each part of the background is shown in the majority of the video, the entire background image needs not to appear in any of its frames. The algorithm is expected to work accurately. == Background image generation == === Equations === For three frames of image sequence x 1 {\displaystyle x_{1}} , x 2 {\displaystyle x_{2}} , and x 3 {\displaystyle x_{3}} , the background image B {\displaystyle B} is obtained using B = x 3 ( x 1 ⊕ x 2 ) + x 1 x 2 {\displaystyle B=x_{3}(x_{1}\oplus x_{2})+x_{1}x_{2}} where ⊕ {\displaystyle \oplus } denotes the exclusive disjunctive bit operator. The Boolean mode function S {\displaystyle S} of the table occurs when the number of 1 entries is larger than half of the number of images such that S = { 1 , if ∑ i = 1 n x i ≥ ⌈ n 2 + 1 ⌉ , and n ≥ 3 0 , otherwise {\displaystyle S={\begin{cases}1,&{\text{if }}\sum _{i=1}^{n}x_{i}\geq \left\lceil {\frac {n}{2}}+1\right\rceil ,{\text{ and }}n\geq 3\\0,&{\text{otherwise}}\end{cases}}} For three images, the background image B {\displaystyle B} can be taken as the value x ¯ 1 x 2 x 3 + x 1 x ¯ 2 x 3 + x 1 x 2 x ¯ 3 + x 1 x 2 x 3 {\displaystyle {\bar {x}}_{1}x_{2}x_{3}+x_{1}{\bar {x}}_{2}x_{3}+x_{1}x_{2}{\bar {x}}_{3}+x_{1}x_{2}x_{3}} === Background generation algorithm === At the first level, three frames are selected at random from the image sequence to produce a background image by combining them using the first equation. This yields a better background image at the second level. The procedure is repeated until desired level L {\displaystyle L} . == Theoretical accuracy == At level ℓ {\displaystyle \ell } , the probability p ℓ {\displaystyle p_{\ell }} that the modal bit predicted is the actual modal bit is represented by the equation p ℓ = ( p ℓ − 1 ) 3 + 3 ( p ℓ − 1 ) 2 ( 1 − p ℓ − 1 ) {\displaystyle p_{\ell }=(p_{\ell -1})^{3}+3(p_{\ell -1})^{2}(1-p_{\ell -1})} . The table below gives the computed probability values across several levels using some specific initial probabilities. It can be observed that even if the modal bit at the considered position is at a low 60% of the frames, the probability of accurate modal bit determination is already more than 99% at 6 levels. == Space complexity == The space requirement of the Teknomo–Fernandez algorithm is given by the function O ( R F + R 3 L ) {\displaystyle O(RF+R3^{L})} , depending on the resolution R {\displaystyle R} of the image, the number F {\displaystyle F} of frames in the video, and the desired number L {\displaystyle L} of levels. However, the fact that L {\displaystyle L} will probably not exceed 6 reduces the space complexity to O ( R F ) {\displaystyle O(RF)} . == Time complexity == The entire algorithm runs in O ( R ) {\displaystyle O(R)} -time, only depending on the resolution of the image. Computing the modal bit for each bit can be done in O ( 1 ) {\displaystyle O(1)} -time while the computation of the resulting image from the three given images can be done in O ( R ) {\displaystyle O(R)} -time. The number of the images to be processed in L {\displaystyle L} levels is O ( 3 L ) {\displaystyle O(3^{L})} . However, since L ≤ 6 {\displaystyle L\leq 6} , then this is actually O ( 1 ) {\displaystyle O(1)} , thus the algorithm runs in O ( R ) {\displaystyle O(R)} . == Variants == A variant of the Teknomo–Fernandez algorithm that incorporates the Monte-Carlo method named CRF has been developed. Two different configurations of CRF were implemented: CRF9,2 and CRF81,1. Experiments on some colored video sequences showed that the CRF configurations outperform the TF algorithm in terms of accuracy. However, the TF algorithm remains more efficient in terms of processing time. == Applications == Object detection Face detection Face recognition Pedestrian detection Video surveillance Motion capture Human-computer interaction Content-based video coding Traffic monitoring Real-time gesture recognition

  Read more →
• 

  Ultra Hal

  Ultra Hal is a chatbot intended to function as a virtual assistant. It was developed by Zabaware, Inc. Ultra Hal uses a natural language interface with animated characters using speech synthesis. Users can communicate with the chatterbot via typing or via a speech recognition engine. It utilizes the WordNet lexical dictionary. Its name is an allusion to HAL 9000, the artificial intelligence from the movie 2001: A Space Odyssey. Ultra Hal won the 2007 Loebner Prize for "most human" chatterbot.

  Read more →