TensorFlow is a software library for machine learning and artificial intelligence. It can be used across a range of tasks, but is used mainly for training and inference of neural networks. It is one of the most popular deep learning frameworks, alongside others such as PyTorch. It is free and open-source software released under the Apache License 2.0. It was developed by the Google Brain team for Google's internal use in research and production. The initial version was released under the Apache License 2.0 in 2015. Google released an updated version, TensorFlow 2.0, in September 2019. TensorFlow can be used in a wide variety of programming languages, including Python, JavaScript, C++, and Java, facilitating its use in a range of applications in many sectors. == History == === DistBelief === Starting in 2011, Google Brain built DistBelief as a proprietary machine learning system based on deep learning neural networks. Its use grew rapidly across diverse Alphabet companies in both research and commercial applications. Google assigned multiple computer scientists, including Jeff Dean, to simplify and refactor the codebase of DistBelief into a faster, more robust application-grade library, which became TensorFlow. In 2009, the team, led by Geoffrey Hinton, had implemented generalized backpropagation and other improvements, which allowed generation of neural networks with substantially higher accuracy, for instance a 25% reduction in errors in speech recognition. === TensorFlow === TensorFlow is Google Brain's second-generation system. Version 1.0.0 was released on February 11, 2017. While the reference implementation runs on single devices, TensorFlow can run on multiple CPUs and GPUs (with optional CUDA and SYCL extensions for general-purpose computing on graphics processing units). TensorFlow is available on 64-bit Linux, macOS, Windows, and mobile computing platforms including Android and iOS. Its flexible architecture allows for easy deployment of computation across a variety of platforms (CPUs, GPUs, TPUs), and from desktops to clusters of servers to mobile and edge devices. TensorFlow computations are expressed as stateful dataflow graphs. The name TensorFlow derives from the operations that such neural networks perform on multidimensional data arrays, which are referred to as tensors. During the Google I/O Conference in June 2016, Jeff Dean stated that 1,500 repositories on GitHub mentioned TensorFlow, of which only 5 were from Google. In March 2018, Google announced TensorFlow.js version 1.0 for machine learning in JavaScript. In Jan 2019, Google announced TensorFlow 2.0. It became officially available in September 2019. In May 2019, Google announced TensorFlow Graphics for deep learning in computer graphics. === Tensor processing unit (TPU) === In May 2016, Google announced its Tensor processing unit (TPU), an application-specific integrated circuit (ASIC, a hardware chip) built specifically for machine learning and tailored for TensorFlow. A TPU is a programmable AI accelerator designed to provide high throughput of low-precision arithmetic (e.g., 8-bit), and oriented toward using or running models rather than training them. Google announced they had been running TPUs inside their data centers for more than a year, and had found them to deliver an order of magnitude better-optimized performance per watt for machine learning. In May 2017, Google announced the second-generation, as well as the availability of the TPUs in Google Compute Engine. The second-generation TPUs deliver up to 180 teraflops of performance, and when organized into clusters of 64 TPUs, provide up to 11.5 petaflops. In May 2018, Google announced the third-generation TPUs delivering up to 420 teraflops of performance and 128 GB high bandwidth memory (HBM). Cloud TPU v3 Pods offer 100+ petaflops of performance and 32 TB HBM. In February 2018, Google announced that they were making TPUs available in beta on the Google Cloud Platform. === Edge TPU === In July 2018, the Edge TPU was announced. Edge TPU is Google's purpose-built ASIC chip designed to run TensorFlow Lite machine learning (ML) models on small client computing devices such as smartphones known as edge computing. === TensorFlow Lite === In May 2017, Google announced TensorFlow Lite as a software stack to support machine learning models for mobile and embedded devices, and in November 2017, provided the developer preview. In January 2019, the TensorFlow team released a developer preview of the mobile GPU inference engine with OpenGL ES 3.1 Compute Shaders on Android devices and Metal Compute Shaders on iOS devices. In May 2019, Google announced that their TensorFlow Lite Micro (also known as TensorFlow Lite for Microcontrollers) and ARM's uTensor would be merging. It was renamed as LiteRT in 2024. === TensorFlow 2.0 === As TensorFlow's market share among research papers was declining to the advantage of PyTorch, the TensorFlow Team announced a release of a new major version of the library in September 2019. TensorFlow 2.0 introduced many changes, the most significant being TensorFlow eager, which changed the automatic differentiation scheme from the static computational graph to the "Define-by-Run" scheme originally made popular by Chainer and later PyTorch. Other major changes included removal of old libraries, cross-compatibility between trained models on different versions of TensorFlow, and significant improvements to the performance on GPU. == Features == === AutoDifferentiation === AutoDifferentiation is the process of automatically calculating the gradient vector of a model with respect to each of its parameters. With this feature, TensorFlow can automatically compute the gradients for the parameters in a model, which is useful to algorithms such as backpropagation which require gradients to optimize performance. To do so, the framework must keep track of the order of operations done to the input Tensors in a model, and then compute the gradients with respect to the appropriate parameters. === Eager execution === TensorFlow includes an "eager execution" mode, which means that operations are evaluated immediately as opposed to being added to a computational graph which is executed later. Code executed eagerly can be examined step-by step-through a debugger, since data is augmented at each line of code rather than later in a computational graph. This execution paradigm is considered to be easier to debug because of its step by step transparency. === Distribute === In both eager and graph executions, TensorFlow provides an API for distributing computation across multiple devices with various distribution strategies. This distributed computing can often speed up the execution of training and evaluating of TensorFlow models and is a common practice in the field of AI. === Losses === To train and assess models, TensorFlow provides a set of loss functions (also known as cost functions). Some popular examples include mean squared error (MSE) and binary cross entropy (BCE). === Metrics === In order to assess the performance of machine learning models, TensorFlow gives API access to commonly used metrics. Examples include various accuracy metrics (binary, categorical, sparse categorical) along with other metrics such as Precision, Recall, and Intersection-over-Union (IoU). === TF.nn === TensorFlow.nn is a module for executing primitive neural network operations on models. Some of these operations include variations of convolutions (1/2/3D, Atrous, depthwise), activation functions (Softmax, RELU, GELU, Sigmoid, etc.) and their variations, and other operations (max-pooling, bias-add, etc.). === Optimizers === TensorFlow offers a set of optimizers for training neural networks, including ADAM, ADAGRAD, and Stochastic Gradient Descent (SGD). When training a model, different optimizers offer different modes of parameter tuning, often affecting a model's convergence and performance. == Usage and extensions == === TensorFlow === TensorFlow serves as a core platform and library for machine learning. TensorFlow's APIs use Keras to allow users to make their own machine-learning models. In addition to building and training their model, TensorFlow can also help load the data to train the model, and deploy it using TensorFlow Serving. TensorFlow provides a stable Python Application Program Interface (API), as well as APIs without backwards compatibility guarantee for JavaScript, C++, and Java. Third-party language binding packages are also available for C#, Haskell, Julia, MATLAB, Object Pascal, R, Scala, Rust, OCaml, and Crystal. Bindings that are now archived and unsupported include Go and Swift. === TensorFlow.js === TensorFlow also has a library for machine learning in JavaScript. Using the provided JavaScript APIs, TensorFlow.js allows users to use either Tensorflow.js models or converted models from TensorFlow or TFLite, retrain the given models, and run on the web. === LiteRT === LiteRT, formerly known as Te
Metadatabase
Metadatabase is a database model for (1) metadata management, (2) global query of independent databases, and (3) distributed data processing. The word metadatabase is an addition to the dictionary. Originally, metadata was only a common term referring simply to "data about data", such as tags, keywords, and markup headers. However, in this technology, the concept of metadata is extended to also include such data and knowledge representation as information models (e.g., relations, entities-relationships, and objects), application logic (e.g., production rules), and analytic models (e.g., simulation, optimization, and mathematical algorithms). In the case of analytic models, it is also referred to as a Modelbase. These classes of metadata are integrated with some modeling ontology to give rise to a stable set of meta-relations (tables of metadata). Individual models are interpreted as metadata and entered into these tables. As such, models are inserted, retrieved, updated, and deleted in the same manner as ordinary data do in an ordinary (relational) database. Users will also formulate global queries and requests for processing of local databases through the Metadatabase, using the globally integrated metadata. The Metadatabase structure can be implemented in any open technology for relational databases. == Significance == The Metadatabase technology is developed at Rensselaer Polytechnic Institute at Troy, New York, by a group of faculty and students (see the references at the end of the article), starting in late 1980s. Its main contribution includes the extension of the concept of metadata and metadata management, and the original approach of designing a database for metadata applications. These conceptual results continue to motivate new research and new applications. At the level of particular design, its openness and scalability is tied to that of the particular ontology proposed: It requires reverse-representation of the application models in order to save them into the meta-relations. In theory, the ontology is neutral, and it has been proven in some industrial applications. However, it needs more development to establish it for the field as an open technology. The requirement of reverse-representation is common to any global information integration technology. A way to facilitate it in the Metadatabase approach is to distribute a core portion of it at each local site, to allow for peer-to-peer translation on the fly.
Data product
In data management and product management, a data product is a reusable, active, and standardized data asset designed to deliver measurable value to its users, whether internal or external, by applying the rigorous principles of product thinking and management. It comprises one or more data artifacts (e.g., datasets, models, pipelines) and is enriched with metadata, including governance policies, data quality rules, data contracts, and, where applicable, a software bill of materials (SBOM) to document its dependencies and components. Ownership of a data product is aligned to a specific domain or use case, ensuring accountability, stewardship, and its continuous evolution throughout its lifecycle. Adhering to the FAIR principles – findable, accessible, interoperable, and reusable – a data product is designed to be discoverable, scalable, reusable, and aligned with both business and regulatory standards, driving innovation and efficiency in modern data ecosystems. == History == In 2012, DJ Patil proposed the first documented definition: a data product is a product that facilitates an end goal through the use of data. In 2019, Zhamak Dehghani introduced Data Mesh, with a strong focus on domain-oriented data products. Later, in 2020, she solidifies Data Mesh around four principles, one being Data as a Product, in which she defines Data Product as the node on the mesh that encapsulates three structural components required for its function, providing access to the domain's analytical data as a product. In 2024, Andrea Gioia published one of the first books specifically on data products post Data Mesh announcement. In his book, Gioia defines the concept of pure data product. In 2025, during the Data Day Texas conference, Jean-Georges Perrin and a collective of product managers and data engineers got together to craft the current definition and make it available to the public domain. In July 2025, Bitol, a project of The Linux Foundation, released and early version of the Open Data Product Standard (ODPS) aiming at normalizing data products
Cryptovirology
Cryptovirology refers to the study of cryptography use in malware, such as ransomware and asymmetric backdoors. Traditionally, cryptography and its applications are defensive in nature, and provide privacy, authentication, and security to users. Cryptovirology employs a twist on cryptography, showing that it can also be used offensively. It can be used to mount extortion based attacks that cause loss of access to information, loss of confidentiality, and information leakage, tasks which cryptography typically prevents. The field was born with the observation that public-key cryptography can be used to break the symmetry between what an antivirus analyst sees regarding malware and what the attacker sees. The antivirus analyst sees a public key contained in the malware, whereas the attacker sees the public key contained in the malware as well as the corresponding private key (outside the malware) since the attacker created the key pair for the attack. The public key allows the malware to perform trapdoor one-way operations on the victim's computer that only the attacker can undo. == Overview == The field encompasses covert malware attacks in which the attacker securely steals private information such as symmetric keys, private keys, PRNG state, and the victim's data. Examples of such covert attacks are asymmetric backdoors. An asymmetric backdoor is a backdoor (e.g., in a cryptosystem) that can be used only by the attacker, even after it is found. This contrasts with the traditional backdoor that is symmetric, i.e., anyone that finds it can use it. Kleptography, a subfield of cryptovirology, is the study of asymmetric backdoors in key generation algorithms, digital signature algorithms, key exchanges, pseudorandom number generators, encryption algorithms, and other cryptographic algorithms. The NIST Dual EC DRBG random bit generator has an asymmetric backdoor in it. The EC-DRBG algorithm utilizes the discrete-log kleptogram from kleptography, which by definition makes the EC-DRBG a cryptotrojan. Like ransomware, the EC-DRBG cryptotrojan contains and uses the attacker's public key to attack the host system. The cryptographer Ari Juels indicated that NSA effectively orchestrated a kleptographic attack on users of the Dual EC DRBG pseudorandom number generation algorithm and that, although security professionals and developers have been testing and implementing kleptographic attacks since 1996, "you would be hard-pressed to find one in actual use until now." Due to public outcry about this cryptovirology attack, NIST rescinded the EC-DRBG algorithm from the NIST SP 800-90 standard. Covert information leakage attacks carried out by cryptoviruses, cryptotrojans, and cryptoworms that, by definition, contain and use the public key of the attacker is a major theme in cryptovirology. In "deniable password snatching," a cryptovirus installs a cryptotrojan that asymmetrically encrypts host data and covertly broadcasts it. This makes it available to everyone, noticeable by no one (except the attacker), and only decipherable by the attacker. An attacker caught installing the cryptotrojan claims to be a virus victim. An attacker observed receiving the covert asymmetric broadcast is one of the thousands, if not millions of receivers, and exhibits no identifying information whatsoever. The cryptovirology attack achieves "end-to-end deniability." It is a covert asymmetric broadcast of the victim's data. Cryptovirology also encompasses the use of private information retrieval (PIR) to allow cryptoviruses to search for and steal host data without revealing the data searched for even when the cryptotrojan is under constant surveillance. By definition, such a cryptovirus carries within its own coding sequence the query of the attacker and the necessary PIR logic to apply the query to host systems. == History == The first cryptovirology attack and discussion of the concept was by Adam L. Young and Moti Yung, at the time called "cryptoviral extortion" and it was presented at the 1996 IEEE Security & Privacy conference. In this attack, a cryptovirus, cryptoworm, or cryptotrojan contains the public key of the attacker and hybrid encrypts the victim's files. The malware prompts the user to send the asymmetric ciphertext to the attacker who will decipher it and return the symmetric decryption key it contains for a fee. The victim needs the symmetric key to decrypt the encrypted files if there is no way to recover the original files (e.g., from backups). The 1996 IEEE paper predicted that cryptoviral extortion attackers would one day demand e-money, long before Bitcoin even existed. Many years later, the media relabeled cryptoviral extortion as ransomware. In 2016, cryptovirology attacks on healthcare providers reached epidemic levels, prompting the U.S. Department of Health and Human Services to issue a Fact Sheet on Ransomware and HIPAA. The fact sheet states that when electronic protected health information is encrypted by ransomware, a breach has occurred, and the attack therefore constitutes a disclosure that is not permitted under HIPAA, the rationale being that an adversary has taken control of the information. Sensitive data might never leave the victim organization, but the break-in may have allowed data to be sent out undetected. California enacted a law that defines the introduction of ransomware into a computer system with the intent of extortion as being against the law. == Examples == === Tremor virus === While viruses in the wild have used cryptography in the past, the only purpose of such usage of cryptography was to avoid detection by antivirus software. For example, the tremor virus used polymorphism as a defensive technique in an attempt to avoid detection by anti-virus software. Though cryptography does assist in such cases to enhance the longevity of a virus, the capabilities of cryptography are not used in the payload. The One-half virus was amongst the first viruses known to have encrypted affected files. === Tro_Ransom.A virus === An example of a virus that informs the owner of the infected machine to pay a ransom is the virus nicknamed Tro_Ransom.A. This virus asks the owner of the infected machine to send $10.99 to a given account through Western Union. Virus.Win32.Gpcode.ag is a classic cryptovirus. This virus partially uses a version of 660-bit RSA and encrypts files with many different extensions. It instructs the owner of the machine to email a given mail ID if the owner desires the decryptor. If contacted by email, the user will be asked to pay a certain amount as ransom in return for the decryptor. === CAPI === It has been demonstrated that using just 8 different calls to Microsoft's Cryptographic API (CAPI), a cryptovirus can satisfy all its encryption needs. == Other uses of cryptography-enabled malware == Apart from cryptoviral extortion, there are other potential uses of cryptoviruses, such as deniable password snatching, cryptocounters, private information retrieval, and in secure communication between different instances of a distributed cryptovirus.
Internet
The Internet (or internet) is the global system of interconnected computer networks that uses the Internet protocol suite (TCP/IP) to communicate between networks and devices. It is a network of networks that comprises private, public, academic, business, and government networks of local to global scope, linked by electronic, wireless, and optical networking technologies. The Internet carries a vast range of information services and resources, such as the interlinked hypertext documents and applications of the World Wide Web (WWW), electronic mail, discussion groups, internet telephony, streaming media and file sharing. Most traditional communication media, including telephone, radio, television, paper mail, newspapers, and print publishing, have been transformed by the Internet, giving rise to new media such as email, online music, digital newspapers, news aggregators, and audio and video streaming websites. The Internet has enabled and accelerated new forms of personal interaction through instant messaging, Internet forums, and social networking services. Online shopping has also grown to occupy a significant market across industries, enabling firms to extend brick and mortar presences to serve larger markets. Business-to-business and financial services on the Internet affect supply chains across entire industries. The origins of the Internet date back to research that enabled the time-sharing of computer resources, the development of packet switching, and the design of computer networks for data communication. The set of communication protocols to enable internetworking on the Internet arose from research and development commissioned in the 1970s by the Defense Advanced Research Projects Agency (DARPA) of the United States Department of Defense in collaboration with universities and researchers across the United States, United Kingdom and France. The Internet has no single centralized governance in either technological implementation or policies for access and usage. Each constituent network sets its own policies. The overarching definitions of the two principal name spaces on the Internet, the Internet Protocol address (IP address) space and the Domain Name System (DNS), are directed by a maintainer organization, the Internet Corporation for Assigned Names and Numbers (ICANN). The technical underpinning and standardization of the core protocols is an activity of the non-profit Internet Engineering Task Force (IETF). == Terminology == The word internetted was used as early as 1849, meaning interconnected or interwoven. The word Internet was used in 1945 by the United States War Department in a radio operator's manual, and in 1974 as the shorthand form of Internetwork. Today, the term Internet most commonly refers to the global system of interconnected computer networks, though it may also refer to any group of smaller networks. The word Internet may be capitalized as a proper noun, although this is becoming less common. This reflects the tendency in English to capitalize new terms and move them to lowercase as they become familiar. The word is sometimes still capitalized to distinguish the global internet from smaller networks, though many publications, including the AP Stylebook since 2016, recommend the lowercase form in every case. In 2016, the Oxford English Dictionary found that, based on a study of around 2.5 billion printed and online sources, "Internet" was capitalized in 54% of cases. The terms Internet and World Wide Web are often used interchangeably; it is common to speak of "going on the Internet" when using a web browser to view web pages. However, the World Wide Web, or the Web, is only one of a large number of Internet services. It is the global collection of web pages, documents and other web resources linked by hyperlinks and URLs. == History == === 1960s === In the 1960s, computer scientists began developing systems for time-sharing of computer resources. J. C. R. Licklider proposed the idea of a universal network while working at Bolt Beranek & Newman and, later, leading the Information Processing Techniques Office at the Advanced Research Projects Agency (ARPA) of the United States Department of Defense. Research into packet switching, one of the fundamental Internet technologies, started in the work of Paul Baran at RAND in the early 1960s and, independently, Donald Davies at the United Kingdom's National Physical Laboratory in 1965. After the Symposium on Operating Systems Principles in 1967, packet switching from the proposed NPL network was incorporated into the design of the ARPANET, an experimental resource sharing network proposed by ARPA. ARPANET development began with two network nodes which were interconnected between the University of California, Los Angeles and the Stanford Research Institute on 29 October 1969. The third site was at the University of California, Santa Barbara, followed by the University of Utah. === 1970s === By the end of 1971, 15 sites were connected to the young ARPANET. Thereafter, the ARPANET gradually developed into a decentralized communications network, connecting remote centers and military bases in the United States. Other user networks and research networks, such as the Merit Network and CYCLADES, were developed in the late 1960s and early 1970s. Early international collaborations for the ARPANET were rare. Connections were made in 1973 to Norway (NORSAR and, later, NDRE) and to Peter Kirstein's research group at University College London, which provided a gateway to British academic networks, the first internetwork for resource sharing. ARPA projects, the International Network Working Group and commercial initiatives led to the development of various protocols and standards by which multiple separate networks could become a single network, or a network of networks. In 1974, Vint Cerf at Stanford University and Bob Kahn at DARPA published a proposal for "A Protocol for Packet Network Intercommunication". Cerf and his graduate students used the term internet as a shorthand for internetwork in RFC 675. The Internet Experiment Notes and later RFCs repeated this use. The work of Louis Pouzin and Robert Metcalfe had important influences on the resulting TCP/IP design. National PTTs and commercial providers developed the X.25 standard and deployed it on public data networks. === 1980s === The ARPANET initially served as a backbone for the interconnection of regional academic and military networks in the United States to enable resource sharing. Access to the ARPANET was expanded in 1981 when the National Science Foundation (NSF) funded the Computer Science Network (CSNET). In 1982, the Internet Protocol Suite (TCP/IP) was standardized, which facilitated worldwide proliferation of interconnected networks. TCP/IP network access expanded again in 1986 when the National Science Foundation Network (NSFNet) provided access to supercomputer sites in the United States for researchers, first at speeds of 56 kbit/s and later at 1.5 Mbit/s and 45 Mbit/s. The NSFNet expanded into academic and research organizations in Europe, Australia, New Zealand and Japan in 1988–89. Although other network protocols such as UUCP and PTT public data networks had global reach well before this time, this marked the beginning of the Internet as an intercontinental network. Commercial Internet service providers emerged in 1989 in the United States and Australia. The ARPANET was decommissioned in 1990. === 1990s === The linking of commercial networks and enterprises by the early 1990s, as well as the advent of the World Wide Web, marked the beginning of the transition to the modern Internet. Steady advances in semiconductor technology and optical networking created new economic opportunities for commercial involvement in the expansion of the network in its core and for delivering services to the public. In mid-1989, MCI Mail and Compuserve established connections to the Internet, delivering email and public access products to the half million users of the Internet. Just months later, on 1 January 1990, PSInet launched an alternate Internet backbone for commercial use; one of the networks that added to the core of the commercial Internet of later years. In March 1990, the first high-speed T1 (1.5 Mbit/s) link between the NSFNET and Europe was installed between Cornell University and CERN, allowing much more robust communications than were capable with satellites. Later in 1990, Tim Berners-Lee began writing WorldWideWeb, the first web browser, after two years of lobbying CERN management. By Christmas 1990, Berners-Lee had built all the tools necessary for a working Web: the HyperText Transfer Protocol (HTTP) 0.9, the HyperText Markup Language (HTML), the first Web browser (which was also an HTML editor and could access Usenet newsgroups and FTP files), the first HTTP server software (later known as CERN httpd), the first web server, and the first Web pages that described the project itself. In 1991 the
Meta-learning (computer science)
Meta-learning is a subfield of machine learning where automatic learning algorithms are applied to metadata about machine learning experiments. As of 2017, the term had not found a standard interpretation, however the main goal is to use such metadata to understand how automatic learning can become flexible in solving learning problems, hence to improve the performance of existing learning algorithms or to learn (induce) the learning algorithm itself, hence the alternative term learning to learn. Flexibility is important because each learning algorithm is based on a set of assumptions about the data, its inductive bias. This means that it will only learn well if the bias matches the learning problem. A learning algorithm may perform very well in one domain, but not on the next. This poses strong restrictions on the use of machine learning or data mining techniques, since the relationship between the learning problem (often some kind of database) and the effectiveness of different learning algorithms is not yet understood. By using different kinds of metadata, like properties of the learning problem, algorithm properties (like performance measures), or patterns previously derived from the data, it is possible to learn, select, alter or combine different learning algorithms to effectively solve a given learning problem. Critiques of meta-learning approaches bear a strong resemblance to the critique of metaheuristic, a possibly related problem. A good analogy to meta-learning, and the inspiration for Jürgen Schmidhuber's early work (1987) and Yoshua Bengio et al.'s work (1991), considers that genetic evolution learns the learning procedure encoded in genes and executed in each individual's brain. In an open-ended hierarchical meta-learning system using genetic programming, better evolutionary methods can be learned by meta evolution, which itself can be improved by meta meta evolution, etc. == Definition == A proposed definition for a meta-learning system combines three requirements: The system must include a learning subsystem. Experience is gained by exploiting meta knowledge extracted in a previous learning episode on a single dataset, or from different domains. Learning bias must be chosen dynamically. Bias refers to the assumptions that influence the choice of explanatory hypotheses and not the notion of bias represented in the bias-variance dilemma. Meta-learning is concerned with two aspects of learning bias. Declarative bias specifies the representation of the space of hypotheses, and affects the size of the search space (e.g., represent hypotheses using linear functions only). Procedural bias imposes constraints on the ordering of the inductive hypotheses (e.g., preferring smaller hypotheses). == Common approaches == There are three common approaches: using (cyclic) networks with external or internal memory (model-based) learning effective distance metrics (metrics-based) explicitly optimizing model parameters for fast learning (optimization-based). === Model-Based === Model-based meta-learning models updates its parameters rapidly with a few training steps, which can be achieved by its internal architecture or controlled by another meta-learner model. ==== Memory-Augmented Neural Networks ==== A Memory-Augmented Neural Network, or MANN for short, is claimed to be able to encode new information quickly and thus to adapt to new tasks after only a few examples. ==== Meta Networks ==== Meta Networks (MetaNet) learns a meta-level knowledge across tasks and shifts its inductive biases via fast parameterization for rapid generalization. === Metric-Based === The core idea in metric-based meta-learning is similar to nearest neighbors algorithms, which weight is generated by a kernel function. It aims to learn a metric or distance function over objects. The notion of a good metric is problem-dependent. It should represent the relationship between inputs in the task space and facilitate problem solving. ==== Convolutional Siamese Neural Network ==== Siamese neural network is composed of two twin networks whose output is jointly trained. There is a function above to learn the relationship between input data sample pairs. The two networks are the same, sharing the same weight and network parameters. ==== Matching Networks ==== Matching Networks learn a network that maps a small labelled support set and an unlabelled example to its label, obviating the need for fine-tuning to adapt to new class types. ==== Relation Network ==== The Relation Network (RN), is trained end-to-end from scratch. During meta-learning, it learns to learn a deep distance metric to compare a small number of images within episodes, each of which is designed to simulate the few-shot setting. ==== Prototypical Networks ==== Prototypical Networks learn a metric space in which classification can be performed by computing distances to prototype representations of each class. Compared to recent approaches for few-shot learning, they reflect a simpler inductive bias that is beneficial in this limited-data regime, and achieve satisfied results. === Optimization-Based === What optimization-based meta-learning algorithms intend for is to adjust the optimization algorithm so that the model can be good at learning with a few examples. ==== LSTM Meta-Learner ==== LSTM-based meta-learner is to learn the exact optimization algorithm used to train another learner neural network classifier in the few-shot regime. The parametrization allows it to learn appropriate parameter updates specifically for the scenario where a set amount of updates will be made, while also learning a general initialization of the learner (classifier) network that allows for quick convergence of training. ==== Temporal Discreteness ==== Model-Agnostic Meta-Learning (MAML) is a fairly general optimization algorithm, compatible with any model that learns through gradient descent. ==== Reptile ==== Reptile is a remarkably simple meta-learning optimization algorithm, given that both of its components rely on meta-optimization through gradient descent and both are model-agnostic. == Examples == Some approaches which have been viewed as instances of meta-learning: Recurrent neural networks (RNNs) are universal computers. In 1993, Jürgen Schmidhuber showed how "self-referential" RNNs can in principle learn by backpropagation to run their own weight change algorithm, which may be quite different from backpropagation. In 2001, Sepp Hochreiter & A.S. Younger & P.R. Conwell built a successful supervised meta-learner based on Long short-term memory RNNs. It learned through backpropagation a learning algorithm for quadratic functions that is much faster than backpropagation. Researchers at Deepmind (Marcin Andrychowicz et al.) extended this approach to optimization in 2017. In the 1990s, Meta Reinforcement Learning or Meta RL was achieved in Schmidhuber's research group through self-modifying policies written in a universal programming language that contains special instructions for changing the policy itself. There is a single lifelong trial. The goal of the RL agent is to maximize reward. It learns to accelerate reward intake by continually improving its own learning algorithm which is part of the "self-referential" policy. An extreme type of Meta Reinforcement Learning is embodied by the Gödel machine, a theoretical construct which can inspect and modify any part of its own software which also contains a general theorem prover. It can achieve recursive self-improvement in a provably optimal way. Model-Agnostic Meta-Learning (MAML) was introduced in 2017 by Chelsea Finn et al. Given a sequence of tasks, the parameters of a given model are trained such that few iterations of gradient descent with few training data from a new task will lead to good generalization performance on that task. MAML "trains the model to be easy to fine-tune." MAML was successfully applied to few-shot image classification benchmarks and to policy-gradient-based reinforcement learning. Variational Bayes-Adaptive Deep RL (VariBAD) was introduced in 2019. While MAML is optimization-based, VariBAD is a model-based method for meta reinforcement learning, and leverages a variational autoencoder to capture the task information in an internal memory, thus conditioning its decision making on the task. When addressing a set of tasks, most meta learning approaches optimize the average score across all tasks. Hence, certain tasks may be sacrificed in favor of the average score, which is often unacceptable in real-world applications. By contrast, Robust Meta Reinforcement Learning (RoML) focuses on improving low-score tasks, increasing robustness to the selection of task. RoML works as a meta-algorithm, as it can be applied on top of other meta learning algorithms (such as MAML and VariBAD) to increase their robustness. It is applicable to both supervised meta learning and meta reinforcement learning. Discovering meta-knowledge works by inducing knowledge
Signatures with efficient protocols
Signatures with efficient protocols are a form of digital signature invented by Jan Camenisch and Anna Lysyanskaya in 2001. In addition to being secure digital signatures, they need to allow for the efficient implementation of two protocols: A protocol for computing a digital signature in a secure two-party computation protocol. A protocol for proving knowledge of a digital signature in a zero-knowledge protocol. In applications, the first protocol allows a signer to possess the signing key to issue a signature to a user (the signature owner) without learning all the messages being signed or the complete signature. The second protocol allows the signature owner to prove that he has a signature on many messages without revealing the signature and only a (possibly) empty subset of the messages. The combination of these two protocols allows for the implementation of digital credential and ecash protocols.