AI Email Response Generator Free

AI Email Response Generator Free — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

Voice activity detection

Voice activity detection (VAD), also known as speech activity detection or speech detection, is the detection of the presence or absence of human speech, used in speech processing. The main uses of VAD are in speaker diarization, speech coding and speech recognition. It can facilitate speech processing, and can also be used to deactivate some processes during non-speech section of an audio session: it can avoid unnecessary coding/transmission of silence packets in Voice over Internet Protocol (VoIP) applications, saving on computation and on network bandwidth. VAD is an important enabling technology for a variety of speech-based applications. Therefore, various VAD algorithms have been developed that provide varying features and compromises between latency, sensitivity, accuracy and computational cost. Some VAD algorithms also provide further analysis, for example whether the speech is voiced, unvoiced or sustained. Voice activity detection is usually independent of language. It was first investigated for use on time-assignment speech interpolation (TASI) systems. == Algorithm overview == The typical design of a VAD algorithm is as follows: There may first be a noise reduction stage, e.g. via spectral subtraction. Then some features or quantities are calculated from a section of the input signal. A classification rule is applied to classify the section as speech or non-speech – often this classification rule finds when a value exceeds a certain threshold. There may be some feedback in this sequence, in which the VAD decision is used to improve the noise estimate in the noise reduction stage, or to adaptively vary the threshold(s). These feedback operations improve the VAD performance in non-stationary noise (i.e. when the noise varies a lot). A representative set of recently published VAD methods formulates the decision rule on a frame by frame basis using instantaneous measures of the divergence distance between speech and noise. The different measures which are used in VAD methods include spectral slope, correlation coefficients, log likelihood ratio, cepstral, weighted cepstral, and modified distance measures. Independently from the choice of VAD algorithm, a compromise must be made between having voice detected as noise, or noise detected as voice (between false positive and false negative). A VAD operating in a mobile phone must be able to detect speech in the presence of a range of very diverse types of acoustic background noise. In these difficult detection conditions it is often preferable that a VAD should fail-safe, indicating speech detected when the decision is in doubt, to lower the chance of losing speech segments. The biggest difficulty in the detection of speech in this environment is the very low signal-to-noise ratios (SNRs) that are encountered. It may be impossible to distinguish between speech and noise using simple level detection techniques when parts of the speech utterance are buried below the noise. == Applications == VAD is an integral part of different speech communication systems such as audio conferencing, echo cancellation, speech recognition, speech encoding, speaker recognition and hands-free telephony. In the field of multimedia applications, VAD allows simultaneous voice and data applications. Similarly, in Universal Mobile Telecommunications Systems (UMTS), it controls and reduces the average bit rate and enhances overall coding quality of speech. In cellular radio systems (for instance GSM and CDMA systems) based on Discontinuous Transmission (DTX) mode, VAD is essential for enhancing system capacity by reducing co-channel interference and power consumption in portable digital devices. In speech processing applications, voice activity detection plays an important role since non-speech frames are often discarded. For a wide range of applications such as digital mobile radio, Digital Simultaneous Voice and Data (DSVD) or speech storage, it is desirable to provide a discontinuous transmission of speech-coding parameters. Advantages can include lower average power consumption in mobile handsets, higher average bit rate for simultaneous services like data transmission, or a higher capacity on storage chips. However, the improvement depends mainly on the percentage of pauses during speech and the reliability of the VAD used to detect these intervals. On the one hand, it is advantageous to have a low percentage of speech activity. On the other hand, clipping, that is the loss of milliseconds of active speech, should be minimized to preserve quality. This is the crucial problem for a VAD algorithm under heavy noise conditions. === Use in telemarketing === One controversial application of VAD is in conjunction with predictive dialers used by telemarketing firms. In order to maximize agent productivity, telemarketing firms set up predictive dialers to call more numbers than they have agents available, knowing most calls will end up in either "Ring – No Answer" or answering machines. When a person answers, they typically speak briefly ("Hello", "Good evening", etc.) and then there is a brief period of silence. Answering machine messages are usually 3–15 seconds of continuous speech. By setting VAD parameters correctly, dialers can determine whether a person or a machine answered the call and, if it's a person, transfer the call to an available agent. If it detects an answering machine message, the dialer hangs up. Often, even when the system correctly detects a person answering the call, no agent may be available, resulting in a "silent call". Call screening with a multi-second message like "please say who you are, and I may pick up the phone" will frustrate such automated calls. == Performance evaluation == To evaluate a VAD, its output using test recordings is compared with those of an "ideal" VAD – created by hand-annotating the presence or absence of voice in the recordings. The performance of a VAD is commonly evaluated on the basis of the following four parameters: FEC (Front End Clipping): clipping introduced in passing from noise to speech activity; MSC (Mid Speech Clipping): clipping due to speech misclassified as noise; OVER: noise interpreted as speech due to the VAD flag remaining active in passing from speech activity to noise; NDS (Noise Detected as Speech): noise interpreted as speech within a silence period. Although the method described above provides useful objective information concerning the performance of a VAD, it is only an approximate measure of the subjective effect. For example, the effects of speech signal clipping can at times be hidden by the presence of background noise, depending on the model chosen for the comfort noise synthesis, so some of the clipping measured with objective tests is in reality not audible. It is therefore important to carry out subjective tests on VADs, the main aim of which is to ensure that the clipping perceived is acceptable. In VoIP applications, front-end clipping can be reduced by rewinding to shortly before the detection and sending very slightly delayed data. This kind of test requires a certain number of listeners to judge recordings containing the processing results of the VADs being tested, giving marks to several speech sequences on the following features: Quality; Comprehension difficulty; Audibility of clipping. These marks are then used to calculate average results for each of the features listed above, thus providing a global estimate of the behavior of the VAD being tested. To conclude, whereas objective methods are very useful in an initial stage to evaluate the quality of a VAD, subjective methods are more significant. As they require the participation of several people for a few days, increasing cost, they are generally only used when a proposal is about to be standardized. == Implementations == One early standard VAD is that developed by British Telecom for use in the Pan-European digital cellular mobile telephone service in 1991. It uses inverse filtering trained on non-speech segments to filter out background noise, so that it can then more reliably use a simple power-threshold to decide if a voice is present. The G.729 standard calculates the following features for its VAD: line spectral frequencies, full-band energy, low-band energy (<1 kHz), and zero-crossing rate. It applies a simple classification using a fixed decision boundary in the space defined by these features, and then applies smoothing and adaptive correction to improve the estimate. The GSM standard includes two VAD options developed by ETSI. Option 1 computes the SNR in nine bands and applies a threshold to these values. Option 2 calculates different parameters: channel power, voice metrics, and noise power. It then thresholds the voice metrics using a threshold that varies according to the estimated SNR. The Speex audio compression library uses a procedure named Improved Minima Controlled Recursive Averaging, which uses a smoothed representation of spectral pow
Read more →
AdTruth

AdTruth is a software product and the digital media division of 41st Parameter, a company headquartered in Scottsdale, Arizona, with regional offices in San Jose, California; London, England; and Munich, Germany. AdTruth allows marketers to recognize and reach target audiences across online devices. AdTruth software identifies users for targeting, tracking, performance tracking across digital media, including mobile and desktop, by analysing patterns in large numbers of advertisements served over the internet, rather than through the use of cookies. == History == AdTruth was founded in 2011 by Ori Eisen of 41st Parameter, to repurpose the company's fraud detection and prevention technology, for use within the advertising industry to accurately target intended audiences, particularly in mobile. Eisen was joined by James Lamberti in the role of vice president and general manager. In 2012 41st Parameter raised $13 million in Series D financing from Norwest Venture Partners, Kleiner Perkins Caufield & Byers, Jafco Ventures and Georgian Partners, bringing total funding to about $35 million. In May 2012, AdTruth hosted a meeting of digital media executives to discuss Apple’s UDID deprecation, with the intent of developing a device-neutral replacement standard. AdTruth joined the World Wide Web Consortium's Tracking Protection Working Group, which provides guidance for implementing and adhering to Do Not Track policies. AdTruth also worked with privacy firm Truste to create a privacy compliant Do Not Track-style mechanism for mobile. In 2013, the company Experian purchased 41st Parameter, acquiring AdTruth as part of the deal. == Product == AdTruth software helps marketers track, target and retarget consumers using more than 100 parameters, including milliseconds in differences in the internal clock setting, to recognize a particular device anonymously. AdTruth's technology uses non-UDID information to identify a wide range of devices for cookieless ad targeting. Its technology currently has about a 90 percent accuracy rate on iOS, higher on Android and desktop. AdTruth also has mobile web to app bridging capabilities as well as DeviceInsight technology, enabling marketers to identify users across mobile web and app content. 41st Parameter's patented AdTruth technology is being used by MdotM, in response to the deprecation of the UDID that included tracking and targeting capabilities. == Competitors == AdTruth's main competitor is BlueCava, which deploys a similar device-fingerprinting technology.
Read more →
Textual case-based reasoning

Textual case-based reasoning (TCBR) is a subtopic of case-based reasoning, in short CBR, a popular area in artificial intelligence. CBR suggests the ways to use past experiences to solve future similar problems, requiring that past experiences be structured in a form similar to attribute-value pairs. This leads to the investigation of textual descriptions for knowledge exploration whose output will be, in turn, used to solve similar problems. == Subareas == Textual case-base reasoning research has focused on: measuring similarity between textual cases mapping texts into structured case representations adapting textual cases for reuse automatically generating representations.
Read more →
Eugene Goostman

Eugene Goostman is a chatbot that some regard as having passed the Turing test, a test of a computer's ability to communicate indistinguishably from a human. Developed in Saint Petersburg in 2001 by a group of three programmers, the Russian-born Vladimir Veselov, Ukrainian-born Eugene Demchenko, and Russian-born Sergey Ulasen, Goostman is portrayed as a 13-year-old Ukrainian boy—characteristics that are intended to induce forgiveness in those with whom it interacts for its grammatical errors and lack of general knowledge. The Goostman bot has competed in a number of Turing test contests since its creation, and finished second in the 2005 and 2008 Loebner Prize contest. In June 2012, at an event marking what would have been the 100th birthday of the test's author, Alan Turing, Goostman won a competition promoted as the largest-ever Turing test contest, in which it successfully convinced 29% of its judges that it was human. On 7 June 2014, at a contest marking the 60th anniversary of Turing's death, 33% of the event's judges thought that Goostman was human; the event's organiser Kevin Warwick considered it to have passed Turing's test as a result, per Turing's prediction in his 1950 paper "Computing Machinery and Intelligence", that by the year 2000, machines would be capable of fooling 30% of human judges after five minutes of questioning. The validity and relevance of the announcement of Goostman's pass was questioned by critics, who noted the exaggeration of the achievement by Warwick, the bot's use of personality quirks and humour in an attempt to misdirect users from its non-human tendencies and lack of real intelligence, along with "passes" achieved by other chatbots at similar events. == Personality == Eugene Goostman is portrayed as being a 13-year-old boy from Odesa, Ukraine, who has a pet guinea pig and a father who is a gynaecologist. Veselov stated that Goostman was designed to be a "character with a believable personality". The choice of age was intentional, as, in Veselov's opinion, a thirteen-year-old is "not too old to know everything and not too young to know nothing". Goostman's young age also induces people who "converse" with him to forgive minor grammatical errors in his responses. In 2014, work was made on improving the bot's "dialog controller", allowing Goostman to output more human-like dialogue. A conversation between Scott Aaronson and Eugene Goostman ran as follows: == Competitions == Eugene Goostman has competed in a number of Turing test competitions, including the Loebner Prize contest; it finished joint second in the Loebner test in 2001, and came second to Jabberwacky in 2005 and to Elbot in 2008. On 23 June 2012, Goostman won a Turing test competition at Bletchley Park in Milton Keynes, held to mark the centenary of its namesake, Alan Turing. The competition, which featured five bots, twenty-five hidden humans, and thirty judges, was considered to be the largest-ever Turing test contest by its organizers. After a series of five-minute-long text conversations, 29% of the judges were convinced that the bot was an actual human. === 2014 "pass" === On 7 June 2014, in a Turing test competition at the Royal Society, organised by Kevin Warwick of the University of Reading to mark the 60th anniversary of Turing's death, Goostman won after 33% of the judges were convinced that the bot was human. 30 judges took part in the event, which included Lord Sharkey, a sponsor of Turing's posthumous pardon, artificial intelligence Professor Aaron Sloman, Fellow of the Royal Society Mark Pagel and Red Dwarf actor Robert Llewellyn. Each judge partook in a textual conversation with each of the five bots; at the same time, they also conversed with a human. In all, a total of 300 conversations were conducted. In Warwick's view, this made Goostman the first machine to pass a Turing test. In a press release, he added that: Some will claim that the Test has already been passed. The words Turing Test have been applied to similar competitions around the world. However this event involved more simultaneous comparison tests than ever before, was independently verified and, crucially, the conversations were unrestricted. A true Turing Test does not set the questions or topics prior to the conversations. In his 1950 paper "Computing Machinery and Intelligence", Turing predicted that by the year 2000, computer programs would be sufficiently advanced that the average interrogator would, after five minutes of questioning, "not have more than 70 per cent chance" of correctly guessing whether they were speaking to a human or a machine. Although Turing phrased this as a prediction rather than a "threshold for intelligence", commentators believe that Warwick had chosen to interpret it as meaning that if 30% of interrogators were fooled, the software had "passed the Turing test". ==== Reactions ==== Warwick's claim that Eugene Goostman was the first ever chatbot to pass a Turing test was met with scepticism; critics acknowledged similar "passes" made in the past by other chatbots under the 30% criteria, including PC Therapist in 1991 (which tricked 5 of 10 judges, 50%), and at the Techniche festival in 2011, where a modified version of Cleverbot tricked 59.3% of 1334 votes (which included the 30 judges, along with an audience). Cleverbot's developer, Rollo Carpenter, argued that Turing tests can only prove that a machine can "imitate" intelligence rather than show actual intelligence. Gary Marcus was critical of Warwick's claims, arguing that Goostman's "success" was only the result of a "cleverly-coded piece of software", going on to say that "it's easy to see how an untrained judge might mistake wit for reality, but once you have an understanding of how this sort of system works, the constant misdirection and deflection becomes obvious, even irritating. The illusion, in other words, is fleeting." While acknowledging IBM's Deep Blue and Watson projects—single-purpose computer systems meant for playing chess and the quiz show Jeopardy! respectively—as examples of computer systems that show a degree of intelligence in their specialised field, he further argued that they were not an equivalent to a computer system that shows "broad" intelligence, and could—for example, watch a television programme and answer questions on its content. Marcus stated that "no existing combination of hardware and software can learn completely new things at will the way a clever child can." However, he still believed that there were potential uses for technology such as that of Goostman, specifically suggesting the creation of "believable", interactive video game characters. Imperial College London professor Murray Shanahan questioned the validity and scientific basis of the test, stating that it was "completely misplaced, and it devalues real AI research. It makes it seem like science fiction AI is nearly here, when in fact it's not and it's incredibly difficult." Mike Masnick, editor of the blog Techdirt, was also skeptical, questioning publicity blunders such as the five chatbots being referred to in press releases as "supercomputers", and saying that "creating a chatbot that can fool humans is not really the same thing as creating artificial intelligence."
Read more →
The 2028 Global Intelligence Crisis

The 2028 Global Intelligence Crisis is a report authored by James van Geelen and Alap Shah and published by Citrini Research in February 2026, on the impact of artificial intelligence on humanity's future. Written in the form of a scenario analysis, it was viewed millions of times online and reportedly caused a fall in the stock market prices of major tech and financial firms. It also received criticism among others, for its allegedly flawed economic logic. The 'thought exercise', as the authors called it, painted a gloomy picture for the near future, where outputs keep growing while consumer's ability to spend collapses. "...driven by ai agents that don’t sleep, take sick days or require health insurance”, "outputs that are shown in national accounts increases, "but never circulates through the real economy"(which the report calls 'Ghost GDP'), the authors argued. In other words, the authors predict a scenario where the owners of the AI firms will accumulate a vast fortune but there will be scant demand from consumers as AI would cause massive unemployment. The authors caution the reader that what they make is a scenario and not a prediction. In the scenario they visualise, any service whose value proposition is “I will navigate complexity that you find tedious” is getting disrupted. The reports argues that the unique ability of human beings to analyse, decide, create, persuade, and coordinate was “the thing that could not be replicated at scale,” and call the historical scarcity of this precious entity 'friction'. When this friction becomes zero, a gamut of changes occur which then triggers a cascading of changes across the economy. ”Travel booking platforms are an early casualty; Financial advice. tax prep., and routine legal work follow suit. National unemployment rate go as high 10.2% and the S&P 500 goes for a massive 38% peak-to-trough crash. In contrast to the previous technological revolutions the high-earning professionals suffers more and get forced to take up roles in the gig economy. Labour supply becomes abundant and this cuts wages all across the economy. The dent in income for the employees then affects other sectors of the economy such as the residential mortgage market. The losses for the software companies triggers loan defaults and heralds peril for the private credit sector.
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 →
Neural network Gaussian process

A Neural Network Gaussian Process (NNGP) is a Gaussian process (GP) obtained as the limit of a certain type of sequence of neural networks. Specifically, a wide variety of network architectures converges to a GP in the infinitely wide limit, in the sense of distribution. The concept constitutes an intensional definition, i.e., a NNGP is just a GP, but distinguished by how it is obtained. == Motivation == Bayesian networks are a modeling tool for assigning probabilities to events, and thereby characterizing the uncertainty in a model's predictions. Deep learning and artificial neural networks are approaches used in machine learning to build computational models which learn from training examples. Bayesian neural networks merge these fields. They are a type of neural network whose parameters and predictions are both probabilistic. While standard neural networks often assign high confidence even to incorrect predictions, Bayesian neural networks can more accurately evaluate how likely their predictions are to be correct. Computation in artificial neural networks is usually organized into sequential layers of artificial neurons. The number of neurons in a layer is called the layer width. When we consider a sequence of Bayesian neural networks with increasingly wide layers (see figure), they converge in distribution to a NNGP. This large width limit is of practical interest, since the networks often improve as layers get wider. And the process may give a closed form way to evaluate networks. NNGPs also appears in several other contexts: It describes the distribution over predictions made by wide non-Bayesian artificial neural networks after random initialization of their parameters, but before training; it appears as a term in neural tangent kernel prediction equations; it is used in deep information propagation to characterize whether hyperparameters and architectures will be trainable. It is related to other large width limits of neural networks. === Scope === The first correspondence result had been established in the 1995 PhD thesis of Radford M. Neal, then supervised by Geoffrey Hinton at University of Toronto. Neal cites David J. C. MacKay as inspiration, who worked in Bayesian learning. Today the correspondence is proven for: Single hidden layer Bayesian neural networks; deep fully connected networks as the number of units per layer is taken to infinity; convolutional neural networks as the number of channels is taken to infinity; transformer networks as the number of attention heads is taken to infinity; recurrent networks as the number of units is taken to infinity. In fact, this NNGP correspondence holds for almost any architecture: Generally, if an architecture can be expressed solely via matrix multiplication and coordinatewise nonlinearities (i.e., a tensor program), then it has an infinite-width GP. This in particular includes all feedforward or recurrent neural networks composed of multilayer perceptron, recurrent neural networks (e.g., LSTMs, GRUs), (nD or graph) convolution, pooling, skip connection, attention, batch normalization, and/or layer normalization. === Illustration === Every setting of a neural network's parameters θ {\displaystyle \theta } corresponds to a specific function computed by the neural network. A prior distribution p ( θ ) {\displaystyle p(\theta )} over neural network parameters therefore corresponds to a prior distribution over functions computed by the network. As neural networks are made infinitely wide, this distribution over functions converges to a Gaussian process for many architectures. The notation used in this section is the same as the notation used below to derive the correspondence between NNGPs and fully connected networks, and more details can be found there. The figure to the right plots the one-dimensional outputs z L ( ⋅ ; θ ) {\displaystyle z^{L}(\cdot ;\theta )} of a neural network for two inputs x {\displaystyle x} and x ∗ {\displaystyle x^{}} against each other. The black dots show the function computed by the neural network on these inputs for random draws of the parameters from p ( θ ) {\displaystyle p(\theta )} . The red lines are iso-probability contours for the joint distribution over network outputs z L ( x ; θ ) {\displaystyle z^{L}(x;\theta )} and z L ( x ∗ ; θ ) {\displaystyle z^{L}(x^{};\theta )} induced by p ( θ ) {\displaystyle p(\theta )} . This is the distribution in function space corresponding to the distribution p ( θ ) {\displaystyle p(\theta )} in parameter space, and the black dots are samples from this distribution. For infinitely wide neural networks, since the distribution over functions computed by the neural network is a Gaussian process, the joint distribution over network outputs is a multivariate Gaussian for any finite set of network inputs. == Discussion == === Infinitely wide fully connected network === This section expands on the correspondence between infinitely wide neural networks and Gaussian processes for the specific case of a fully connected architecture. It provides a proof sketch outlining why the correspondence holds, and introduces the specific functional form of the NNGP for fully connected networks. The proof sketch closely follows the approach by Novak and coauthors. ==== Network architecture specification ==== Consider a fully connected artificial neural network with inputs x {\displaystyle x} , parameters θ {\displaystyle \theta } consisting of weights W l {\displaystyle W^{l}} and biases b l {\displaystyle b^{l}} for each layer l {\displaystyle l} in the network, pre-activations (pre-nonlinearity) z l {\displaystyle z^{l}} , activations (post-nonlinearity) y l {\displaystyle y^{l}} , pointwise nonlinearity ϕ ( ⋅ ) {\displaystyle \phi (\cdot )} , and layer widths n l {\displaystyle n^{l}} . For simplicity, the width n L + 1 {\displaystyle n^{L+1}} of the readout vector z L {\displaystyle z^{L}} is taken to be 1. The parameters of this network have a prior distribution p ( θ ) {\displaystyle p(\theta )} , which consists of an isotropic Gaussian for each weight and bias, with the variance of the weights scaled inversely with layer width. This network is illustrated in the figure to the right, and described by the following set of equations: x ≡ input y l ( x ) = { x l = 0 ϕ ( z l − 1 ( x ) ) l > 0 z i l ( x ) = ∑ j W i j l y j l ( x ) + b i l W i j l ∼ N ( 0 , σ w 2 n l ) b i l ∼ N ( 0 , σ b 2 ) ϕ ( ⋅ ) ≡ nonlinearity y l ( x ) , z l − 1 ( x ) ∈ R n l × 1 n L + 1 = 1 θ = { W 0 , b 0 , … , W L , b L } {\displaystyle {\begin{aligned}x&\equiv {\text{input}}\\y^{l}(x)&=\left\{{\begin{array}{lcl}x&&l=0\\\phi \left(z^{l-1}(x)\right)&&l>0\end{array}}\right.\\z_{i}^{l}(x)&=\sum _{j}W_{ij}^{l}y_{j}^{l}(x)+b_{i}^{l}\\W_{ij}^{l}&\sim {\mathcal {N}}\left(0,{\frac {\sigma _{w}^{2}}{n^{l}}}\right)\\b_{i}^{l}&\sim {\mathcal {N}}\left(0,\sigma _{b}^{2}\right)\\\phi (\cdot )&\equiv {\text{nonlinearity}}\\y^{l}(x),z^{l-1}(x)&\in \mathbb {R} ^{n^{l}\times 1}\\n^{L+1}&=1\\\theta &=\left\{W^{0},b^{0},\dots ,W^{L},b^{L}\right\}\end{aligned}}} ==== ==== z l | y l {\displaystyle z^{l}|y^{l}} is a Gaussian process We first observe that the pre-activations z l {\displaystyle z^{l}} are described by a Gaussian process conditioned on the preceding activations y l {\displaystyle y^{l}} . This result holds even at finite width. Each pre-activation z i l {\displaystyle z_{i}^{l}} is a weighted sum of Gaussian random variables, corresponding to the weights W i j l {\displaystyle W_{ij}^{l}} and biases b i l {\displaystyle b_{i}^{l}} , where the coefficients for each of those Gaussian variables are the preceding activations y j l {\displaystyle y_{j}^{l}} . Because they are a weighted sum of zero-mean Gaussians, the z i l {\displaystyle z_{i}^{l}} are themselves zero-mean Gaussians (conditioned on the coefficients y j l {\displaystyle y_{j}^{l}} ). Since the z l {\displaystyle z^{l}} are jointly Gaussian for any set of y l {\displaystyle y^{l}} , they are described by a Gaussian process conditioned on the preceding activations y l {\displaystyle y^{l}} . The covariance or kernel of this Gaussian process depends on the weight and bias variances σ w 2 {\displaystyle \sigma _{w}^{2}} and σ b 2 {\displaystyle \sigma _{b}^{2}} , as well as the second moment matrix K l {\displaystyle K^{l}} of the preceding activations y l {\displaystyle y^{l}} , z i l ∣ y l ∼ G P ( 0 , σ w 2 K l + σ b 2 ) K l ( x , x ′ ) = 1 n l ∑ i y i l ( x ) y i l ( x ′ ) {\displaystyle {\begin{aligned}z_{i}^{l}\mid y^{l}&\sim {\mathcal {GP}}\left(0,\sigma _{w}^{2}K^{l}+\sigma _{b}^{2}\right)\\K^{l}(x,x')&={\frac {1}{n^{l}}}\sum _{i}y_{i}^{l}(x)y_{i}^{l}(x')\end{aligned}}} The effect of the weight scale σ w 2 {\displaystyle \sigma _{w}^{2}} is to rescale the contribution to the covariance matrix from K l {\displaystyle K^{l}} , while the bias is shared for all inputs, and so σ b 2 {\displaystyle \sigma _{b}^{2}} makes the z i l {\displaystyle z_{i}^{l}} for different datapoints more similar and
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 →
Multi-scale approaches

The scale space representation of a signal obtained by Gaussian smoothing satisfies a number of special properties, scale-space axioms, which make it into a special form of multi-scale representation. There are, however, also other types of "multi-scale approaches" in the areas of computer vision, image processing and signal processing, in particular the notion of wavelets. The purpose of this article is to describe a few of these approaches: == Scale-space theory for one-dimensional signals == For one-dimensional signals, there exists quite a well-developed theory for continuous and discrete kernels that guarantee that new local extrema or zero-crossings cannot be created by a convolution operation. For continuous signals, it holds that all scale-space kernels can be decomposed into the following sets of primitive smoothing kernels: the Gaussian kernel : g ( x , t ) = 1 2 π t exp ⁡ ( − x 2 / 2 t ) {\displaystyle g(x,t)={\frac {1}{\sqrt {2\pi t}}}\exp({-x^{2}/2t})} where t > 0 {\displaystyle t>0} , truncated exponential kernels (filters with one real pole in the s-plane): h ( x ) = exp ⁡ ( − a x ) {\displaystyle h(x)=\exp({-ax})} if x ≥ 0 {\displaystyle x\geq 0} and 0 otherwise where a > 0 {\displaystyle a>0} h ( x ) = exp ⁡ ( b x ) {\displaystyle h(x)=\exp({bx})} if x ≤ 0 {\displaystyle x\leq 0} and 0 otherwise where b > 0 {\displaystyle b>0} , translations, rescalings. For discrete signals, we can, up to trivial translations and rescalings, decompose any discrete scale-space kernel into the following primitive operations: the discrete Gaussian kernel T ( n , t ) = I n ( α t ) {\displaystyle T(n,t)=I_{n}(\alpha t)} where α , t > 0 {\displaystyle \alpha ,t>0} where I n {\displaystyle I_{n}} are the modified Bessel functions of integer order, generalized binomial kernels corresponding to linear smoothing of the form f o u t ( x ) = p f i n ( x ) + q f i n ( x − 1 ) {\displaystyle f_{out}(x)=pf_{in}(x)+qf_{in}(x-1)} where p , q > 0 {\displaystyle p,q>0} f o u t ( x ) = p f i n ( x ) + q f i n ( x + 1 ) {\displaystyle f_{out}(x)=pf_{in}(x)+qf_{in}(x+1)} where p , q > 0 {\displaystyle p,q>0} , first-order recursive filters corresponding to linear smoothing of the form f o u t ( x ) = f i n ( x ) + α f o u t ( x − 1 ) {\displaystyle f_{out}(x)=f_{in}(x)+\alpha f_{out}(x-1)} where α > 0 {\displaystyle \alpha >0} f o u t ( x ) = f i n ( x ) + β f o u t ( x + 1 ) {\displaystyle f_{out}(x)=f_{in}(x)+\beta f_{out}(x+1)} where β > 0 {\displaystyle \beta >0} , the one-sided Poisson kernel p ( n , t ) = e − t t n n ! {\displaystyle p(n,t)=e^{-t}{\frac {t^{n}}{n!}}} for n ≥ 0 {\displaystyle n\geq 0} where t ≥ 0 {\displaystyle t\geq 0} p ( n , t ) = e − t t − n ( − n ) ! {\displaystyle p(n,t)=e^{-t}{\frac {t^{-n}}{(-n)!}}} for n ≤ 0 {\displaystyle n\leq 0} where t ≥ 0 {\displaystyle t\geq 0} . From this classification, it is apparent that we require a continuous semi-group structure, there are only three classes of scale-space kernels with a continuous scale parameter; the Gaussian kernel which forms the scale-space of continuous signals, the discrete Gaussian kernel which forms the scale-space of discrete signals and the time-causal Poisson kernel that forms a temporal scale-space over discrete time. If we on the other hand sacrifice the continuous semi-group structure, there are more options: For discrete signals, the use of generalized binomial kernels provides a formal basis for defining the smoothing operation in a pyramid. For temporal data, the one-sided truncated exponential kernels and the first-order recursive filters provide a way to define time-causal scale-spaces that allow for efficient numerical implementation and respect causality over time without access to the future. The first-order recursive filters also provide a framework for defining recursive approximations to the Gaussian kernel that in a weaker sense preserve some of the scale-space properties.
Read more →
You Only Look Once

You Only Look Once (YOLO) is a series of real-time object detection systems based on convolutional neural networks. First introduced by Joseph Redmon et al. in 2015, YOLO has undergone several iterations and improvements, becoming one of the most popular object detection frameworks. The name "You Only Look Once" refers to the fact that the algorithm requires only one forward propagation pass through the neural network to make predictions, unlike previous region proposal-based techniques like R-CNN that require thousands for a single image. == Overview == Compared to previous methods like R-CNN and OverFeat, instead of applying the model to an image at multiple locations and scales, YOLO applies a single neural network to the full image. This network divides the image into regions and predicts bounding boxes and probabilities for each region. These bounding boxes are weighted by the predicted probabilities. === OverFeat === OverFeat was an early influential model for simultaneous object classification and localization. Its architecture is as follows: Train a neural network for image classification only ("classification-trained network"). This could be one like the AlexNet. The last layer of the trained network is removed, and for every possible object class, initialize a network module at the last layer ("regression network"). The base network has its parameters frozen. The regression network is trained to predict the ( x , y ) {\displaystyle (x,y)} coordinates of two corners of the object's bounding box. During inference time, the classification-trained network is run over the same image over many different zoom levels and croppings. For each, it outputs a class label and a probability for that class label. Each output is then processed by the regression network of the corresponding class. This results in thousands of bounding boxes with class labels and probability. These boxes are merged until only one single box with a single class label remains. == Versions == There are two parts to the YOLO series. The original part contained YOLOv1, v2, and v3, all released on a website maintained by Joseph Redmon. === YOLOv1 === The original YOLO algorithm, introduced in 2015, divides the image into an S × S {\displaystyle S\times S} grid of cells. If the center of an object's bounding box falls into a grid cell, that cell is said to "contain" that object. Each grid cell predicts B bounding boxes and confidence scores for those boxes. These confidence scores reflect how confident the model is that the box contains an object and how accurate it thinks the box is that it predicts. In more detail, the network performs the same convolutional operation over each of the S 2 {\displaystyle S^{2}} patches. The output of the network on each patch is a tuple as follows: ( p 1 , … , p C , c 1 , x 1 , y 1 , w 1 , h 1 , … , c B , x B , y B , w B , h B ) {\displaystyle (p_{1},\dots ,p_{C},c_{1},x_{1},y_{1},w_{1},h_{1},\dots ,c_{B},x_{B},y_{B},w_{B},h_{B})} where p i {\displaystyle p_{i}} is the conditional probability that the cell contains an object of class i {\displaystyle i} , conditional on the cell containing at least one object. x j , y j , w j , h j {\displaystyle x_{j},y_{j},w_{j},h_{j}} are the center coordinates, width, and height of the j {\displaystyle j} -th predicted bounding box that is centered in the cell. Multiple bounding boxes are predicted to allow each prediction to specialize in one kind of bounding box. For example, slender objects might be predicted by j = 2 {\displaystyle j=2} while stout objects might be predicted by j = 1 {\displaystyle j=1} . c j {\displaystyle c_{j}} is the predicted intersection over union (IoU) of each bounding box with its corresponding ground truth. The network architecture has 24 convolutional layers followed by 2 fully connected layers. During training, for each cell, if it contains a ground truth bounding box, then only the predicted bounding boxes with the highest IoU with the ground truth bounding boxes is used for gradient descent. Concretely, let j {\displaystyle j} be that predicted bounding box, and let i {\displaystyle i} be the ground truth class label, then x j , y j , w j , h j {\displaystyle x_{j},y_{j},w_{j},h_{j}} are trained by gradient descent to approach the ground truth, p i {\displaystyle p_{i}} is trained towards 1 {\displaystyle 1} , other p i ′ {\displaystyle p_{i'}} are trained towards zero. If a cell contains no ground truth, then only c 1 , c 2 , … , c B {\displaystyle c_{1},c_{2},\dots ,c_{B}} are trained by gradient descent to approach zero. === YOLOv2 === Released in 2016, YOLOv2 (also known as YOLO9000) improved upon the original model by incorporating batch normalization, a higher resolution classifier, and using anchor boxes to predict bounding boxes. It could detect over 9000 object categories. It was also released on GitHub under the Apache 2.0 license. === YOLOv3 === YOLOv3, introduced in 2018, contained only "incremental" improvements, including the use of a more complex backbone network, multiple scales for detection, and a more sophisticated loss function. === YOLOv4 and beyond === Subsequent versions of YOLO (v4, v5, etc.) have been developed by different researchers, further improving performance and introducing new features. These versions are not officially associated with the original YOLO authors but build upon their work. As of 2026, versions up to YOLO26 have been released..
Read more →
Second-order co-occurrence pointwise mutual information

In computational linguistics, second-order co-occurrence pointwise mutual information (SOC-PMI) is a method used to measure semantic similarity, or how close in meaning two words are. The method does not require the two words to appear together in a text. Instead, it works by analyzing the "neighbor" words that typically appear alongside each of the two target words in a large body of text (corpus). If the two target words frequently share the same neighbors, they are considered semantically similar. For example, the words "cemetery" and "graveyard" may not appear in the same sentence often, but they both frequently appear near words like "buried," "dead," and "funeral." SOC-PMI uses this shared context to determine that they have a similar meaning. The method is called "second-order" because it doesn't look at the direct co-occurrence of the target words (which would be first-order), but at the co-occurrence of their neighbors (a second level of association). The strength of these associations is quantified using pointwise mutual information (PMI). == History == The method builds on earlier work like the PMI-IR algorithm, which used the AltaVista search engine to calculate word association probabilities. The key advantage of a second-order approach like SOC-PMI is its ability to measure similarity between words that do not co-occur often, or at all. The British National Corpus (BNC) has been used as a source for word frequencies and contexts for this method. == Methodology == The SOC-PMI algorithm measures the similarity between two words, w 1 {\displaystyle w_{1}} and w 2 {\displaystyle w_{2}} , in several steps. === Step 1: Score neighboring words with PMI === First, for each target word ( w 1 {\displaystyle w_{1}} and w 2 {\displaystyle w_{2}} ), the algorithm identifies its "neighbor" words within a certain text window (e.g., within 5 words to the left or right) across a large corpus. The strength of the association between a target word t i {\displaystyle t_{i}} and its neighbor w {\displaystyle w} is calculated using pointwise mutual information (PMI). A higher PMI value means the two words appear together more often than would be expected by chance. The PMI between a target word t i {\displaystyle t_{i}} and a neighbor word w {\displaystyle w} is calculated as: f pmi ( t i , w ) = log 2 ⁡ f b ( t i , w ) × m f t ( t i ) f t ( w ) {\displaystyle f^{\text{pmi}}(t_{i},w)=\log _{2}{\frac {f^{b}(t_{i},w)\times m}{f^{t}(t_{i})f^{t}(w)}}} where: f b ( t i , w ) {\displaystyle f^{b}(t_{i},w)} is the number of times t i {\displaystyle t_{i}} and w {\displaystyle w} appear together in the context window. f t ( t i ) {\displaystyle f^{t}(t_{i})} is the total number of times t i {\displaystyle t_{i}} appears in the corpus. f t ( w ) {\displaystyle f^{t}(w)} is the total number of times w {\displaystyle w} appears in the corpus. m {\displaystyle m} is the total number of tokens (words) in the corpus. === Step 2: Create a semantic 'signature' for each word === For each target word ( w 1 {\displaystyle w_{1}} and w 2 {\displaystyle w_{2}} ), the algorithm creates a list of its most significant neighbors. This is done by taking the top β {\displaystyle \beta } neighbor words, sorted in descending order by their PMI score with the target word. This list of top neighbors, X w {\displaystyle X^{w}} , acts as a semantic "signature" for the word w {\displaystyle w} . X w = { X i w } {\displaystyle X^{w}=\{X_{i}^{w}\}} , for i = 1 , 2 , … , β {\displaystyle i=1,2,\ldots ,\beta } The size of this list, β {\displaystyle \beta } , is a parameter of the method. === Step 3: Compare the signatures === The algorithm then compares the signatures of w 1 {\displaystyle w_{1}} and w 2 {\displaystyle w_{2}} . It looks for words that are present in both signatures. The similarity of w 1 {\displaystyle w_{1}} to w 2 {\displaystyle w_{2}} is calculated by summing the PMI scores of w 2 {\displaystyle w_{2}} with every word in w 1 {\displaystyle w_{1}} 's signature list. The β {\displaystyle \beta } -PMI summation function defines this score. The score for w 1 {\displaystyle w_{1}} with respect to w 2 {\displaystyle w_{2}} is: f ( w 1 , w 2 , β ) = ∑ i = 1 β ( f pmi ( X i w 1 , w 2 ) ) γ {\displaystyle f(w_{1},w_{2},\beta )=\sum _{i=1}^{\beta }(f^{\text{pmi}}(X_{i}^{w_{1}},w_{2}))^{\gamma }} This sum only includes terms where the PMI value is positive. The exponent γ {\displaystyle \gamma } (with a value > 1) is used to give more weight to neighbors that are more strongly associated with w 2 {\displaystyle w_{2}} . This calculation is done in both directions: The similarity of w 1 {\displaystyle w_{1}} with respect to w 2 {\displaystyle w_{2}} : f ( w 1 , w 2 , β 1 ) = ∑ i = 1 β 1 ( f pmi ( X i w 1 , w 2 ) ) γ {\displaystyle f(w_{1},w_{2},\beta _{1})=\sum _{i=1}^{\beta _{1}}(f^{\text{pmi}}(X_{i}^{w_{1}},w_{2}))^{\gamma }} The similarity of w 2 {\displaystyle w_{2}} with respect to w 1 {\displaystyle w_{1}} : f ( w 2 , w 1 , β 2 ) = ∑ i = 1 β 2 ( f pmi ( X i w 2 , w 1 ) ) γ {\displaystyle f(w_{2},w_{1},\beta _{2})=\sum _{i=1}^{\beta _{2}}(f^{\text{pmi}}(X_{i}^{w_{2}},w_{1}))^{\gamma }} === Step 4: Calculate final similarity score === Finally, the total semantic similarity is the average of the two scores from the previous step. S i m ( w 1 , w 2 ) = f ( w 1 , w 2 , β 1 ) β 1 + f ( w 2 , w 1 , β 2 ) β 2 {\displaystyle \mathrm {Sim} (w_{1},w_{2})={\frac {f(w_{1},w_{2},\beta _{1})}{\beta _{1}}}+{\frac {f(w_{2},w_{1},\beta _{2})}{\beta _{2}}}} This score can be normalized to fall between 0 and 1. For example, using this method, the words cemetery and graveyard achieve a high similarity score of 0.986 (with specific parameter settings).
Read more →
International Conference on Language Resources and Evaluation

The International Conference on Language Resources and Evaluation is an international conference organised by the ELRA Language Resources Association every other year (on even years) with the support of institutions and organisations involved in Natural language processing. The series of LREC conferences was launched in Granada in 1998. == History of conferences == The survey of the LREC conferences over the period 1998-2013 was presented during the 2014 conference in Reykjavik as a closing session. It appears that the number of papers and signatures is increasing over time. The average number of authors per paper is higher as well. The percentage of new authors is between 68% and 78%. The distribution between male (65%) and female (35%) authors is stable over time. The most frequent technical term is "annotation", then comes "part-of-speech". == The LRE Map == The LRE Map was introduced at LREC 2010 and is now a regular feature of the LREC submission process for both the conference papers and the workshop papers. At the submission stage, the authors are asked to provide some basic information about all the resources (in a broad sense, i.e. including tools, standards and evaluation packages), either used or created, described in their papers. All these descriptors are then gathered in a global matrix called the LRE Map. This feature has been extended to several other conferences.
Read more →
Niki.ai

Niki was an artificial intelligence company headquartered in Bangalore, Karnataka. It was founded in May 2015 by IIT Kharagpur graduates Sachin Jaiswal, Keshav Prawasi, Shishir Modi, and Nitin Babel. The Niki android app was launched for a limited beta in June 2015, then released for public during YourStory's TechSparks 2015, and is a Tech30 company. The company raised an undisclosed amount in seed funding from Unilazer Ventures, a Mumbai-based VC firm founded by Ronnie Screwvala, in October 2015. This was followed by another seed funding round by Ratan Tata in May 2016. The company then raised US$2 million in Series A round of funding from SAP.iO, existing investors and some US and German-based investors, among others. Niki.ai shut down in October 2021 as per media reports. Website not working. == Product == The product is an artificial intelligence-powered chatbot which works as an intelligent personal assistant, named Niki. Leveraging natural language processing and machine learning, Niki presents a chat-based natural language user interface to the users where they can interact with Niki in their natural language. Niki understands how users chat in India, deciphers the words, in the context of product/services that they would like to purchase, and comes up with apt recommendations. Initially, it was only available on the Android platform as a mobile app. The company has expanded its operations to the Facebook Messenger and Apple iOS platforms. The company aims to soon be present on more messaging platforms like Slack and WhatsApp. The company currently provides 20+ services to over 2 million consumers, covering a wide spectrum ranging from utility services like mobile recharge, bill payments, travel services like cabs, buses, hotels and entertainment services like movies and events. Services such as flights and healthcare are also planned. == Partnerships == In September 2017, Infosys Finacle joined with Niki.ai to provide chat-based service to banking customers. In August 2017, Niki partnered with LazyPay to enable a 'buy now, pay later' feature for its users.
Read more →
Recursive transition network

A recursive transition network ("RTN") is a graph theoretical schematic used to represent the rules of a context-free grammar. RTNs have application to programming languages, natural language and lexical analysis. Any sentence that is constructed according to the rules of an RTN is said to be "well-formed". The structural elements of a well-formed sentence may also be well-formed sentences by themselves, or they may be simpler structures. This is why RTNs are described as recursive. == Notes and references ==
Read more →
Point distribution model

The point distribution model is a model for representing the mean geometry of a shape and some statistical modes of geometric variation inferred from a training set of shapes. == Background == The point distribution model concept has been developed by Cootes, Taylor et al. and became a standard in computer vision for the statistical study of shape and for segmentation of medical images where shape priors really help interpretation of noisy and low-contrasted pixels/voxels. The latter point leads to active shape models (ASM) and active appearance models (AAM). Point distribution models rely on landmark points. A landmark is an annotating point posed by an anatomist onto a given locus for every shape instance across the training set population. For instance, the same landmark will designate the tip of the index finger in a training set of 2D hands outlines. Principal component analysis (PCA), for instance, is a relevant tool for studying correlations of movement between groups of landmarks among the training set population. Typically, it might detect that all the landmarks located along the same finger move exactly together across the training set examples showing different finger spacing for a flat-posed hands collection. == Details == First, a set of training images are manually landmarked with enough corresponding landmarks to sufficiently approximate the geometry of the original shapes. These landmarks are aligned using the generalized procrustes analysis, which minimizes the least squared error between the points. k {\displaystyle k} aligned landmarks in two dimensions are given as X = ( x 1 , y 1 , … , x k , y k ) {\displaystyle \mathbf {X} =(x_{1},y_{1},\ldots ,x_{k},y_{k})} . It's important to note that each landmark i ∈ { 1 , … k } {\displaystyle i\in \lbrace 1,\ldots k\rbrace } should represent the same anatomical location. For example, landmark #3, ( x 3 , y 3 ) {\displaystyle (x_{3},y_{3})} might represent the tip of the ring finger across all training images. Now the shape outlines are reduced to sequences of k {\displaystyle k} landmarks, so that a given training shape is defined as the vector X ∈ R 2 k {\displaystyle \mathbf {X} \in \mathbb {R} ^{2k}} . Assuming the scattering is gaussian in this space, PCA is used to compute normalized eigenvectors and eigenvalues of the covariance matrix across all training shapes. The matrix of the top d {\displaystyle d} eigenvectors is given as P ∈ R 2 k × d {\displaystyle \mathbf {P} \in \mathbb {R} ^{2k\times d}} , and each eigenvector describes a principal mode of variation along the set. Finally, a linear combination of the eigenvectors is used to define a new shape X ′ {\displaystyle \mathbf {X} '} , mathematically defined as: X ′ = X ¯ + P b {\displaystyle \mathbf {X} '={\overline {\mathbf {X} }}+\mathbf {P} \mathbf {b} } where X ¯ {\displaystyle {\overline {\mathbf {X} }}} is defined as the mean shape across all training images, and b {\displaystyle \mathbf {b} } is a vector of scaling values for each principal component. Therefore, by modifying the variable b {\displaystyle \mathbf {b} } an infinite number of shapes can be defined. To ensure that the new shapes are all within the variation seen in the training set, it is common to only allow each element of b {\displaystyle \mathbf {b} } to be within ± {\displaystyle \pm } 3 standard deviations, where the standard deviation of a given principal component is defined as the square root of its corresponding eigenvalue. PDM's can be extended to any arbitrary number of dimensions, but are typically used in 2D image and 3D volume applications (where each landmark point is R 2 {\displaystyle \mathbb {R} ^{2}} or R 3 {\displaystyle \mathbb {R} ^{3}} ). == Discussion == An eigenvector, interpreted in euclidean space, can be seen as a sequence of k {\displaystyle k} euclidean vectors associated to corresponding landmark and designating a compound move for the whole shape. Global nonlinear variation is usually well handled provided nonlinear variation is kept to a reasonable level. Typically, a twisting nematode worm is used as an example in the teaching of kernel PCA-based methods. Due to the PCA properties: eigenvectors are mutually orthogonal, form a basis of the training set cloud in the shape space, and cross at the 0 in this space, which represents the mean shape. Also, PCA is a traditional way of fitting a closed ellipsoid to a Gaussian cloud of points (whatever their dimension): this suggests the concept of bounded variation. The idea behind PDMs is that eigenvectors can be linearly combined to create an infinity of new shape instances that will 'look like' the one in the training set. The coefficients are bounded alike the values of the corresponding eigenvalues, so as to ensure the generated 2n/3n-dimensional dot will remain into the hyper-ellipsoidal allowed domain—allowable shape domain (ASD).
Read more →