Big-memory computers are machines with a large amount of random-access memory (RAM). The computers are required for databases, graph analytics, or more generally, high-performance computing, data science, and big data. Some database systems called in-memory databases are designed to run mostly in memory, rarely if ever retrieving data from disk or flash memory. See list of in-memory databases. == Details == The performance of big-memory systems depends on how the central processing units (CPUs) access the memory, via a conventional memory controller or via non-uniform memory access (NUMA). Performance also depends on the size and design of the CPU cache. Performance also depends on operating system (OS) design. The huge pages feature in Linux and other OSes can improve the efficiency of virtual memory. The transparent huge pages feature in Linux can offer better performance for some big-memory workloads. The "Large-Page Support" in Microsoft Windows enables server applications to establish large-page memory regions which are typically three orders of magnitude larger than the native page size.
Quantum robotics
Quantum robotics is an interdisciplinary field that investigates the intersection of robotics and quantum mechanics. This field, in particular, explores the applications of quantum phenomena such as quantum entanglement within the realm of robotics. Examples of its applications include quantum communication in multi-agent cooperative robotic scenarios, the use of quantum algorithms in performing robotics tasks, and the integration of quantum devices (e.g., quantum detectors) in robotic systems. == Introduction == The free-space quantum communication between mobile platforms was proposed for reconfigurable quantum key distribution (QKD) applications using unmanned aerial vehicle (UAVs, a.k.a. drones) in 2017. This technology was later advanced in various aspects in mobile drone and vehicle platforms in several configurations such as drone-to-drone, drone-to-moving vehicle, and vehicle-to-vehicle systems. Some research has contributed to low-size, low-weight, and low-power quantum key distribution systems for small-form UAVs, the characterization of a polarization-based receiver for mobile free-space optical QKD, and optical-relayed entanglement distribution using drones as mobile nodes. The topic of free-space quantum communication between mobile platforms, initially developed to meet the need for free-space QKD and entanglement distribution using mobile nodes, was brought into the robotics domain as an emerging interdisciplinary mechatronics topic to investigate the interface between quantum technologies and the robotic systems domain. The main advantage of such integrated technology is the guaranteed security in communication between multi-agent and cooperative autonomous systems. Other advances are anticipated. == Quantum entanglement == According to quantum mechanics, entanglement occurs when more than one particle become connected. If the state of one particle changes then it will instantly change the state of other particles regardless of their distance. Entangled sensors do the same kind of work and achieve strong sensitivity. A group of quantum robots can measure magnetic fields, gravitational fields and other physical properties using entangled sensors with high rate of accuracy. Again the connection of one robot to other is increased (become strong) by quantum entanglement. == Quantum teleportation == Quantum teleportation is the transfer of quantum information (not physical objects). This is used in case of multi robot process. One robot is programmed with a complex quantum update. Then that robot can teleport that complex quantum information (the update) to other robots. This teleportation or communication is very secure because all the work is done in quantum state. == Kinematics == Quantum computing has been proposed as being optimal for calculating inverse kinematics values. == Alice and Bob robots == In the realm of quantum mechanics, the names Alice and Bob are frequently employed to illustrate various phenomena, protocols, and applications. These include their roles in QKD, quantum cryptography, entanglement, and teleportation. The terms "Alice Robot" and "Bob Robot" serve as analogous expressions that merge the concepts of Alice and Bob from quantum mechanics with mechatronic mobile platforms (such as robots, drones, and autonomous vehicles). For example, the Alice Robot functions as a transmitter platform that communicates with the Bob Robot, housing the receiving detectors.
Rider Spoke
Rider Spoke developed by Blast Theory in collaboration with the Mixed Reality Lab was first staged at the Barbican, London in October 2007. Created for cyclists, it combines elements of theatre, performance, game play and state of the art technology. Rider Spoke was built in the IPerG project on the EQUIP architecture. Rider Spoke has since been presented in Athens (2008), Brighton (2008), Budapest (2008), Sydney (2009, Adelaide (2009) and Liverpool (2010).
Futel
Futel is a public arts organization in Portland, Oregon dedicated to preserving and maintaining public telephone hardware and offering free phone and basic information services. Futel was founded by Karl Anderson, a former software engineer, and Elijah St. Clair. == Technology == Karl Anderson stated that one motivation for the project was to explore the idea of urban furniture. Other reasons were to preserve an important part of hacker history, and to salvage and re-use manufactured items at the end of their lifecycle. The original Futel phones were set up in Portland, Oregon. The organization cleans and repurposes old public payphones which are often salvaged from Craigslist or scrappers. Using interface boxes, they are converted into VoIP phones which are made available publicly, with no cost for phone calls. Anderson has said the service runs on "Asterisk and OpenVPN and a lot of scripts." The payphones operate using publicly-available internet connections. The phones have automated phone trees and users can make a call to local social services, to a weather forecast line, or access local transit information. Volunteers act as telephone operators, offering information about the Futel service, or are available for conversation. Users using Futel's phones may also access voicemail boxes. The system has a "wildcard line" where people can listen to samples of audio left on the main voicemail line along with commentary from Anderson and others. == Network == In February 2021, there were 10 Futel phones in Portland and 3 in other cities. Phones were set up in Detroit and Ypsilanti, Michigan, and Long Beach, Washington. The organization has provided free phone service for a Portland-area homeless encampment after receiving funding from the Awesome Foundation. In 2019 the organization reported their phones being used to make 12,000 phone calls. Futel also said their usage went up and not down during the first year of the COVID-19 pandemic when they outfitted their phone kiosks with handwashing stations and used volunteers to keep the phones clean. The project is funded is primarily through grants and is staffed with volunteers. The project has inspired others such as the PhilTel project in Philadelphia and the RandTel project in Randolph, Vermont. Futel publishes a zine called Party Line.
Hardware trojan
A hardware trojan (HT) is a malicious modification of the circuitry of an integrated circuit. A hardware trojan is completely characterized by its physical representation and its behavior. The payload of an HT is the entire activity that the Trojan executes when it is triggered. In general, trojans try to bypass or disable the security fence of a system: for example, leaking confidential information by radio emission. HTs also could disable, damage or destroy the entire chip or components of it. Hardware trojans may be introduced as hidden front-doors that are inserted while designing a computer chip, by using a pre-made application-specific integrated circuit (ASIC) semiconductor intellectual property core (IP core) that have been purchased from a non-reputable source, or inserted internally by a rogue employee, either acting on their own, or on behalf of rogue special interest groups, or state sponsored spying and espionage. One paper published by IEEE in 2015 explains how a hardware design containing a trojan could leak a cryptographic key leaked over an antenna or network connection, provided that the correct "easter egg" trigger is applied to activate the data leak. In high security governmental IT departments, hardware trojans are a well known problem when buying hardware such as: a KVM switch, keyboards, mice, network cards, or other network equipment. This is especially the case when purchasing such equipment from non-reputable sources that could have placed hardware trojans to leak keyboard passwords, or provide remote unauthorized entry. == Background == In a diverse global economy, outsourcing of production tasks is a common way to lower a product's cost. Embedded hardware devices are not always produced by the firms that design and/or sell them, nor in the same country where they will be used. Outsourced manufacturing can raise doubt about the evidence for the integrity of the manufactured product (i.e., one's certainty that the end-product has no design modifications compared to its original design). Anyone with access to the manufacturing process could, in theory, introduce some change to the final product. For complex products, small changes with large effects can be difficult to detect. The threat of a serious, malicious, design alteration can be especially relevant to government agencies. Resolving doubt about hardware integrity is one way to reduce technology vulnerabilities in the military, finance, energy and political sectors of an economy. Since fabrication of integrated circuits in untrustworthy factories is common, advanced detection techniques have emerged to discover when an adversary has hidden additional components in, or otherwise sabotaged, the circuit's function. == Characterization of hardware trojans == An HT can be characterized by several methods such as by its physical representation, activation phase and its action phase. Alternative methods characterize the HT by trigger, payload and stealth. === Physical characteristics === One of this physical trojan characteristics is the type. The type of a trojan can be either functional or parametric. A trojan is functional if the adversary adds or deletes any transistors or gates to the original chip design. The other kind of trojan, the parametric trojan, modifies the original circuitry, e.g. thinning of wires, weakening of flip-flops or transistors, subjecting the chip to radiation, or using focused ion-beams (FIB) to reduce the reliability of a chip. The size of a trojan is its physical extension or the number of components it is made of. Because a trojan can consist of many components, the designer can distribute the parts of a malicious logic on the chip. The additional logic can occupy the chip wherever it is needed to modify, add, or remove a function. Malicious components can be scattered, called loose distribution, or consist of only few components, called tight distribution, so the area is small where the malicious logic occupies the layout of the chip. In some cases, high-effort adversaries in may regenerate the layout so that the placement of the components of the IC is altered. In rare cases the chip dimension is altered. These changes are structural alterations. === Activation characteristics === The typical trojan is condition-based: It is triggered by sensors, internal logic states, a particular input pattern or an internal counter value. Condition-based trojans are detectable with power traces to some degree when inactive. That is due to the leakage currents generated by the trigger or counter circuit activating the trojan. Hardware trojans can be triggered in different ways. A trojan can be internally activated, which means it monitors one or more signals inside the IC. The malicious circuitry could wait for a count down logic an attacker added to the chip, so that the trojan awakes after a specific time-span. The opposite is externally activated. There can be malicious logic inside a chip, that uses an antenna or other sensors the adversary can reach from outside the chip. For example, a trojan could be inside the control system of a cruising missile. The owner of the missile does not know, that the enemy will be able to switch off the rockets by radio. A trojan which is always-on can be a reduced wire. A chip that is modified in this way produces errors or fails every time the wire is used intensely. Always-on circuits are hard to detect with power trace. In this context combinational trojans and sequential trojans are distinguished. A combinational trojan monitors internal signals until a specific condition happens. A sequential trojan is also an internally activated condition-based circuit, but it monitors the internal signals and searches for sequences not for a specific state or condition like the combinational trojans do. ==== Cryptographic key extraction ==== Extraction of secret keys by means of a hardware trojan without detecting the trojan requires that the trojan uses a random signal or some cryptographic implementation itself. To avoid storing a cryptographic key in the trojan itself and reduction, a physical unclonable function can be used. Physical unclonable functions are small in size and can have an identical layout while the cryptographic properties are different. === Action characteristics === A HT could modify the chip's function or could change the chip's parametric properties (e.g. provokes a process delay). Confidential information can also be transmitted to the adversary (transmission of key information). === Peripheral device hardware trojans === A relatively new threat vector to networks and network endpoints is a HT appearing as a physical peripheral device that is designed to interact with the network endpoint using the approved peripheral device's communication protocol. For example, a USB keyboard that hides all malicious processing cycles from the target network endpoint to which it is attached by communicating with the target network endpoint using unintended USB channels. Once sensitive data is ex-filtrated from the target network endpoint to the HT, the HT can process the data and decide what to do with the data: store the data to memory for later physical retrieval of the HT or possibly ex-filtrate the data to the internet using wireless or using the compromised network endpoint as a pivot. == Potential of threat == A common trojan is passive most of the time-span an altered device is in use. If a trojan is activated the device functionality can be changed, the device can be destroyed or disabled, the device can leak confidential information or the HT may tear down the security and safety of the device. Trojans are stealthy, to avoid detection of the trojan the precondition for activation is a very rare event. Traditional testing techniques are not sufficient. A manufacturing fault happens at a random position while malicious changes are well placed to avoid detection. == Detection == === Physical inspection === First, the molding coat is cut to reveal the circuitry. Then, the engineer repeatedly scans the surface while grinding the layers of the chip. There are several operations to scan the circuitry. Typical visual inspection methods are: scanning optical microscopy (SOM), scanning electron microscopy (SEM), pico-second imaging circuit analysis (PICA), voltage contrast imaging (VCI), light induced voltage alteration (LIVA) or charge induced voltage alteration (CIVA). To compare the floor plan of the chip has to be compared with the image of the actual chip. This is still quite challenging to do. To detect Trojan hardware which include (crypto) keys which are different, an image diff can be taken to reveal the different structure on the chip. The only known hardware Trojan using unique crypto keys but having the same structure is. This property enhances the undetectability of the trojan. === Functional testing === This detection method stimulates the input ports of a chip and monitors the output
Retrieval-augmented generation
Retrieval-augmented generation (RAG) is a technique that enables large language models (LLMs) to retrieve and incorporate new information from external data sources. With RAG, LLMs first refer to a specified set of documents, then respond to user queries. These documents supplement information from the LLM's pre-existing training data. This allows LLMs to use domain-specific and/or updated information that is not available in the training data. For example, this enables LLM-based chatbots to access internal company data or generate responses based on authoritative sources. RAG improves LLMs by incorporating information retrieval before generating responses. Unlike LLMs that rely on static training data, RAG pulls relevant text from databases, uploaded documents, or web sources. According to Ars Technica, "RAG is a way of improving LLM performance, in essence by blending the LLM process with a web search or other document look-up process to help LLMs stick to the facts." This method helps reduce AI hallucinations, which have caused chatbots to describe policies that don't exist, or recommend nonexistent legal cases to lawyers that are looking for citations to support their arguments. RAG also reduces the need to retrain LLMs with new data, saving on computational and financial costs. Beyond efficiency gains, RAG also allows LLMs to include sources in their responses, so users can verify the cited sources. This provides greater transparency, as users can cross-check retrieved content to ensure accuracy and relevance. The term retrieval-augmented generation (RAG) was introduced in a 2020 paper that described combining a parametric language model with a non-parametric external memory accessed through retrieval at inference time. == RAG and LLM limitations == LLMs can provide incorrect information. For example, when Google first demonstrated its LLM tool "Google Bard" (later re-branded to Gemini), the LLM provided incorrect information about the James Webb Space Telescope. This error contributed to a $100 billion decline in Google's stock value. RAG is used to prevent these errors, but it does not solve all the problems. For example, LLMs can generate misinformation even when pulling from factually correct sources if they misinterpret the context. MIT Technology Review gives the example of an AI-generated response stating, "The United States has had one Muslim president, Barack Hussein Obama." The model retrieved this from an academic book rhetorically titled Barack Hussein Obama: America's First Muslim President? The LLM did not "know" or "understand" the context of the title, generating a false statement. LLMs with RAG are programmed to prioritize new information. This technique has been called "prompt stuffing." Without prompt stuffing, the LLM's input is generated by a user; with prompt stuffing, additional relevant context is added to this input to guide the model's response. This approach provides the LLM with key information early in the prompt, encouraging it to prioritize the supplied data over pre-existing training knowledge. == Process == Retrieval-augmented generation (RAG) enhances large language models (LLMs) by incorporating an information-retrieval mechanism that allows models to access and utilize additional data beyond their original training set. Ars Technica notes that "when new information becomes available, rather than having to retrain the model, all that's needed is to augment the model's external knowledge base with the updated information" ("augmentation"). IBM states that "in the generative phase, the LLM draws from the augmented prompt and its internal representation of its training data to synthesize" an answer. === RAG key stages === Typically, the data to be referenced is converted into LLM embeddings, numerical representations in the form of a large vector space. RAG can be used on unstructured (usually text), semi-structured, or structured data (for example knowledge graphs). These embeddings are then stored in a vector database to allow for document retrieval. Given a user query, a document retriever is first called to select the most relevant documents that will be used to augment the query. This comparison can be done using a variety of methods, which depend in part on the type of indexing used. The model feeds this relevant retrieved information into the LLM via prompt engineering of the user's original query. Newer implementations (as of 2023) can also incorporate specific augmentation modules with abilities such as expanding queries into multiple domains and using memory and self-improvement to learn from previous retrievals. Finally, the LLM can generate output based on both the query and the retrieved documents. Some models incorporate extra steps to improve output, such as the re-ranking of retrieved information, context selection, and fine-tuning. == Applications == Retrieval-augmented generation is used in applications where generated responses need to be grounded in external or frequently updated information. Commonly cited use cases include search engines, question-answering systems, customer support chatbots, enterprise knowledge assistants, content generation, recommendation systems, retail and e-commerce, and industrial or manufacturing workflows. In healthcare, RAG has been studied as a way to ground large language model outputs in external medical knowledge sources, although reviews have noted continuing challenges around evaluation, ethics, and clinical reliability. == Improvements == Improvements to the basic process above can be applied at different stages in the RAG flow. === Encoder === These methods focus on the encoding of text as either dense or sparse vectors. Sparse vectors, which encode the identity of a word, are typically dictionary-length and contain mostly zeros. Dense vectors, which encode meaning, are more compact and contain fewer zeros. Various enhancements can improve the way similarities are calculated in the vector stores (databases). Performance improves by optimizing how vector similarities are calculated. Dot products enhance similarity scoring, while approximate nearest neighbor (ANN) searches improve retrieval efficiency over K-nearest neighbors (KNN) searches. Accuracy may be improved with Late Interactions, which allow the system to compare words more precisely after retrieval. This helps refine document ranking and improve search relevance. Hybrid vector approaches may be used to combine dense vector representations with sparse one-hot vectors, taking advantage of the computational efficiency of sparse dot products over dense vector operations. Other retrieval techniques focus on improving accuracy by refining how documents are selected. Some retrieval methods combine sparse representations, such as SPLADE, with query expansion strategies to improve search accuracy and recall. === Retriever-centric methods === These methods aim to enhance the quality of document retrieval in vector databases: Pre-training the retriever using the Inverse Cloze Task (ICT), a technique that helps the model learn retrieval patterns by predicting masked text within documents. Supervised retriever optimization aligns retrieval probabilities with the generator model's likelihood distribution. This involves retrieving the top-k vectors for a given prompt, scoring the generated response's perplexity, and minimizing KL divergence between the retriever's selections and the model's likelihoods to refine retrieval. Reranking techniques can refine retriever performance by prioritizing the most relevant retrieved documents during training. === Language model === By redesigning the language model with the retriever in mind, a 25-time smaller network can get comparable perplexity as its much larger counterparts. Because it is trained from scratch, this method (Retro) incurs the high cost of training runs that the original RAG scheme avoided. The hypothesis is that by giving domain knowledge during training, Retro needs less focus on the domain and can devote its smaller weight resources only to language semantics. The redesigned language model is shown here. It has been reported that Retro is not reproducible, so modifications were made to make it so. The more reproducible version is called Retro++ and includes in-context RAG. === Chunking === Chunking involves various strategies for breaking up the data into vectors so the retriever can find details in it. Three types of chunking strategies are: Fixed length with overlap. This is fast and easy. Overlapping consecutive chunks helps to maintain semantic context across chunks. Syntax-based chunks can break the document up into sentences. Libraries such as spaCy or NLTK can also help. File format-based chunking. Certain file types have natural chunks built in, and it's best to respect them. For example, code files are best chunked and vectorized as whole functions or classes. HTML files should leave