In cryptography, a pepper is a secret added to an input such as a password during hashing with a cryptographic hash function. This value differs from a salt in that it is not stored alongside a password hash, but rather the pepper is kept separate using another meachanism, such as a Hardware Security Module. Note that the National Institute of Standards and Technology refers to this value as a secret key rather than a pepper. A pepper is similar in concept to a salt or an encryption key. It is like a salt in that it is a randomized value that is added to a password hash, and it is similar to an encryption key in that it should be kept secret. A pepper performs a comparable role to a salt or an encryption key, but while a salt is not secret (merely unique) and can be stored alongside the hashed output, a pepper is secret and must not be stored with the output. The hash and salt are usually stored in a database, but, if stored, a pepper must be stored separately to prevent it from being obtained by the attacker in case of a database breach. == History == The idea of a site- or service-specific salt (in addition to a per-user salt) has a long history, with Steven M. Bellovin proposing a local parameter in a Bugtraq post in 1995. In 1996 Udi Manber also described the advantages of such a scheme, terming it a secret salt. However, he suggested not storing the value of the secret salt, but instead rediscovering it by trial and error at password verification time. The term pepper has been used, by analogy to salt, but with a variety of meanings. For example, when discussing a challenge-response scheme, pepper has been used for a salt-like quantity, though not used for password storage; it has been used for a data transmission technique where a pepper must be guessed; and even as a part of jokes. The term pepper was proposed for a secret or local parameter stored separately from the password in a discussion of protecting passwords from rainbow table attacks. This usage did not immediately catch on: for example, Fred Wenzel added support to Django password hashing for storage based on a combination of bcrypt and HMAC with separately stored nonces, without using the term. Usage has since become more common. == Types == There are multiple different types of pepper: A shared secret that is common to all users. A randomly-selected number that must be re-discovered on every password input. These mechanisms could be combined with password salting, iterated hashing or even one another. == Shared-secret pepper == Bellovin and Webster suggest prepend a shared secret to the password before hashing, which allows easy use of existing hash functions. For example, consider two users to be added to a database. This table contains two combinations of username and password. The password is not saved, and the 8-byte (64-bit) 44534C70C6883DE2 pepper is saved in a safe place separate from the output values of the hash, in this case SHA256. Unlike the salt, the pepper does not provide protection to users who use the same password, but protects against dictionary attacks, unless the attacker has the pepper value available. Since the same pepper is not shared between different applications, an attacker is unable to reuse the hashes of one compromised database to another. A complete scheme for saving passwords may include both salt and pepper use. For example, it has been suggested to combine the pepper by encrypting salted password hashes, which allows rotation of the pepper. In the case of a shared-secret pepper, a single compromised password (via password reuse or other attack) along with a user's salt can lead to an attack to discover the pepper, rendering it ineffective. If an attacker knows a plaintext password and a user's salt, as well as the algorithm used to hash the password, then discovering the pepper can be a matter of brute forcing the values of the pepper. This is why NIST recommends the secret value be at least 112 bits, so that discovering it by exhaustive search is prohibitively expensive. The pepper must be generated anew for every application it is deployed in, otherwise a breach of one application would result in lowered security of another application. Without knowledge of the pepper, other passwords in the database will be far more difficult to extract from their hashed values, as the attacker would need to guess the password as well as the pepper. A pepper adds security to a database of salts and hashes because unless the attacker is able to obtain the pepper, cracking even a single hash is intractable, no matter how weak the original password. Even with a list of (salt, hash) pairs, an attacker must also guess the secret pepper in order to find the password which produces the hash. The NIST specification for a secret salt suggests using a Password-Based Key Derivation Function (PBKDF) with an approved Pseudorandom Function such as HMAC with SHA-3 as the hash function of the HMAC. The NIST recommendation is also to perform at least 1000 iterations of the PBKDF, and a further minimum 1000 iterations using the secret salt in place of the non-secret salt. == Randomly-selected pepper that must be re-discovered == The aim of this mechanism is to slow down password the password verification step, thus slowing attacks. The aim is similar increasing the iteration count on bcrypt or Argon2, but the mechanism is different. The secret salt or pepper must be rediscovered by the verifier or attacker each time by guessing. In this situation, the password hashing function is calculated using both the password and the pepper. At password storage time, the pepper is chosen randomly from a range between 1 and R, the hash output is calculated using the password and the pepper. The hash output is stored with the username. The pepper is then discarded. At password verification time, the verifier is provided with a username and password to verify. The originally calculated hash is retrieved for the given username, and then the hash of the password and each value between 1 and R is calculated. If any of these hash values match the stored password hash, the password is considered valid. Note, the possible values of the pepper should not be tested in a fixed order known to an attacker, otherwise a timing attack may reveal the pepper. If the password is correct, the correct pepper will be found in R/2 hash evaluations on average. If the password is incorrect, all R values must be tested before the password can be rejected.
Quantum machine learning
Quantum machine learning (QML) is the study of quantum algorithms for machine learning. It often refers to quantum algorithms for machine learning tasks which analyze classical data, sometimes called quantum-enhanced machine learning. QML algorithms use qubits and quantum operations to try to improve the space and time complexity of classical machine learning algorithms. Hybrid QML methods involve both classical and quantum processing, where computationally difficult subroutines are outsourced to a quantum device. These routines can be more complex in nature and executed faster on a quantum computer. Furthermore, quantum algorithms can be used to analyze quantum states instead of classical data. The term "quantum machine learning" is sometimes used to refer classical machine learning methods applied to data generated from quantum experiments (i.e. machine learning of quantum systems), such as learning the phase transitions of a quantum system or creating new quantum experiments. QML also extends to a branch of research that explores methodological and structural similarities between certain physical systems and learning systems, in particular neural networks. For example, some mathematical and numerical techniques from quantum physics are applicable to classical deep learning and vice versa. Furthermore, researchers investigate more abstract notions of learning theory with respect to quantum information, sometimes referred to as "quantum learning theory". == Machine learning with quantum computers == Quantum-enhanced machine learning refers to quantum algorithms that solve tasks in machine learning, thereby improving and often expediting classical machine learning techniques. Such algorithms typically require one to encode the given classical data set into a quantum computer to make it accessible for quantum information processing. Subsequently, quantum information processing routines are applied and the result of the quantum computation is read out by measuring the quantum system. For example, the outcome of the measurement of a qubit reveals the result of a binary classification task. While many proposals of QML algorithms are still purely theoretical and require a full-scale universal quantum computer to be tested, others have been implemented on small-scale or special purpose quantum devices. === Quantum associative memories and quantum pattern recognition === Early work on quantum associative memories has been done by Dan Ventura and Tony Martinez and by Carlo A. Trugenberger in the late 1990s and early 2000s. Associative (or content-addressable) memories are able to recognize stored content on the basis of a similarity measure, while random access memories are accessed by the address of stored information and not its content. As such they must be able to retrieve both incomplete and corrupted patterns, the essential machine learning task of pattern recognition. Typical classical associative memories store p patterns in the O ( n 2 ) {\displaystyle O(n^{2})} interactions (synapses) of a real, symmetric energy matrix over a network of n artificial neurons. The encoding is such that the desired patterns are local minima of the energy functional and retrieval is done by minimizing the total energy, starting from an initial configuration. Unfortunately, classical associative memories are severely limited by the phenomenon of cross-talk. When too many patterns are stored, spurious memories appear which quickly proliferate, so that the energy landscape becomes disordered and no retrieval is anymore possible. The number of storable patterns is typically limited by a linear function of the number of neurons, p ≤ O ( n ) {\displaystyle p\leq O(n)} . Quantum associative memories (in their simplest realization) store patterns in a unitary matrix U acting on the Hilbert space of n qubits. Retrieval is realized by the unitary evolution of a fixed initial state to a quantum superposition of the desired patterns with probability distribution peaked on the most similar pattern to an input. By its very quantum nature, the retrieval process is thus probabilistic. Because quantum associative memories are free from cross-talk, however, spurious memories are never generated. Correspondingly, they have a superior capacity than classical ones. The number of parameters in the unitary matrix U is O ( p n ) {\displaystyle O(pn)} . One can thus have efficient, spurious-memory-free quantum associative memories for any polynomial number of patterns. If the matrix U is encoded as a unique operator (as opposed as to a sequence of gates as in the circuit model), e.g. by an optical interferometer, the retrieval becomes efficient even for an exponential number of patterns. === Linear algebra simulation with quantum amplitudes === A number of quantum algorithms for machine learning are based on the idea of amplitude encoding, that is, to associate the amplitudes of a quantum state with the inputs and outputs of computations. Since a state of n {\displaystyle n} qubits is described by 2 n {\displaystyle 2^{n}} complex amplitudes, this information encoding can allow for an exponentially compact representation. Intuitively, this corresponds to associating a discrete probability distribution over binary random variables with a classical vector. The goal of algorithms based on amplitude encoding is to formulate quantum algorithms whose resources grow polynomially in the number of qubits n {\displaystyle n} , which amounts to a logarithmic time complexity in the number of amplitudes and thereby the dimension of the input. Many QML algorithms in this category are based on variations of the quantum algorithm for linear systems of equations (colloquially called HHL, after the paper's authors) which, under specific conditions, performs a matrix inversion using an amount of physical resources growing only logarithmically in the dimensions of the matrix. One of these conditions is that a Hamiltonian which entry-wise corresponds to the matrix can be simulated efficiently, which is known to be possible if the matrix is sparse or low rank. For reference, any known classical algorithm for matrix inversion requires a number of operations that grows more than quadratically in the dimension of the matrix (e.g. O ( n 2.373 ) {\displaystyle O{\mathord {\left(n^{2.373}\right)}}} ), but they are not restricted to sparse matrices. Quantum matrix inversion can be applied to machine learning methods in which the training reduces to solving a linear system of equations, for example in least-squares linear regression, the least-squares version of support vector machines, and Gaussian processes. A crucial bottleneck of methods that simulate linear algebra computations with the amplitudes of quantum states is state preparation, which often requires one to initialise a quantum system in a state whose amplitudes reflect the features of the entire dataset. Although efficient methods for state preparation are known for specific cases, this step easily hides the complexity of the task. === Variational quantum algorithms (VQAs) === In a variational quantum algorithm, a classical computer optimizes the parameters used to prepare a quantum state, while a quantum computer is used to do the actual state preparation and measurement. VQAs are considered promising candidates for noisy intermediate-scale quantum computers. Variational quantum circuits (or parameterized quantum circuits) are a popular class of VQAs where the parameters are those used in a fixed quantum circuit. Researchers have studied VQCs to solve optimization problems and find the ground state energy of complex quantum systems, which were difficult to solve using a classical computer. === Quantum binary classifier === Pattern reorganization is one of the important tasks of machine learning, binary classification is one of the tools or algorithms to find patterns. Binary classification is used in supervised learning and in unsupervised learning. In QML, classical bits are converted to qubits and they are mapped to Hilbert space; complex value data are used in a quantum binary classifier to use the advantage of Hilbert space. By exploiting the quantum mechanic properties such as superposition, entanglement, interference the quantum binary classifier produces the accurate result in short period of time. === Quantum machine learning algorithms based on Grover search === Another approach to improving classical machine learning with quantum information processing uses amplitude amplification methods based on Grover's search algorithm, which has been shown to solve unstructured search problems with a quadratic speedup compared to classical algorithms. These quantum routines can be employed for learning algorithms that translate into an unstructured search task, as can be done, for instance, in the case of the k-medians and the k-nearest neighbors algorithms. Other applications include quadratic speedups in the training of perceptrons. An e
Integreat
Integreat (former project name: Refguide+) is an open source mobile app that provides local information and services tailored to refugees and migrants coming to Germany. The content is maintained by local organizations, such as local governments or integration officers, and made available in locally relevant languages. It was developed by Tür an Tür - Digitalfabrik gGmbH (formerly Tür an Tür - Digital Factory gGmbH) in Augsburg together with a team of researchers and students from the Technical University of Munich. == History == In 1997, the Augsburg association "Tür an Tür", which has been working for refugees since 1992, published the brochure "First Steps", which answers local everyday questions. Since addresses and contact persons change quickly, some information is already outdated after a few weeks. Students of business informatics at the Technical University of Munich therefore developed the app Integreat within eight months together with the association and the social department of the city of Augsburg. The app was then also used by other cities and districts within months. As of February 3, 2022, information is available at 72 locations, including Munich, Dortmund, Nuremberg and Augsburg. == Mode of action == Refugees need information on areas such as registration, contact persons, health care, education, family, work and everyday life. Integreat seeks to provide refugees with this information by allowing them to select their geographic location and receive locally relevant information. This information is available offline once the app is opened so it can be used without an internet connection. In addition, the content is translated into the native languages of refugees and migrants to facilitate access. The content is licensed with a CC BY 4.0 license to facilitate collaboration and translation between content creators and dissemination of the content. Integreat is now being used for a broader migrant audience and says it can also support professionals, volunteers, and counseling centers. == Comparable mobile apps == Other mobile apps that are likewise intended to provide initial orientation for refugees include the app Ankommen, a joint project of the Federal Office for Migration and Refugees, the Goethe-Institut, the Federal Employment Agency and the Bavarian Broadcasting Corporation, which is intended as a companion for the first few weeks in Germany, and the Welcome App, a company-sponsored non-profit initiative for information about Germany and asylum procedures with a regional focus, and a book by the Konrad Adenauer Foundation (KAS) and Verlag Herder with a corresponding app Deutschland - Erste Informationen für Flüchtlinge (Germany - First Information for Refugees) as a companion for Arabic-speaking refugees in Germany.
Natural language processing
Natural language processing (NLP) is the processing of natural language information by a computer. NLP is a subfield of computer science and is closely associated with artificial intelligence. NLP is also related to information retrieval, knowledge representation, computational linguistics, and linguistics more broadly. Major processing tasks in an NLP system include: speech recognition, text classification, natural language understanding, and natural language generation. == History == Natural language processing has its roots in the 1950s. Already in 1950, Alan Turing published an article titled "Computing Machinery and Intelligence," which proposed what is now called the Turing test as a criterion of intelligence, though at the time that was not articulated as a problem separate from artificial intelligence. The proposed test includes a task that involves the automated interpretation and generation of natural language. === Symbolic NLP (1950s – early 1990s) === The premise of symbolic NLP is often illustrated using John Searle's Chinese room thought experiment: Given a collection of rules (e.g., a Chinese phrasebook, with questions and matching answers), the computer emulates natural language understanding (or other NLP tasks) by applying those rules to the data it confronts. 1950s: The Georgetown experiment in 1954 involved fully automatic translation of more than sixty Russian sentences into English. The authors claimed that within three or five years, machine translation would be a solved problem. However, real progress was much slower, and after the ALPAC report in 1966, which found that ten years of research had failed to fulfill the expectations, funding for machine translation was dramatically reduced. Little further research in machine translation was conducted in America (though some research continued elsewhere, such as Japan and Europe) until the late 1980s when the first statistical machine translation systems were developed. 1960s: Some notably successful natural language processing systems developed in the 1960s were SHRDLU, a natural language system working in restricted "blocks worlds" with restricted vocabularies, and ELIZA, a simulation of Rogerian psychotherapy, written by Joseph Weizenbaum between 1964 and 1966. Despite using minimal information about human thought or emotion, ELIZA was able to produce interactions that appeared human-like. When the "patient" exceeded the very small knowledge base, ELIZA might provide a generic response, for example, responding to "My head hurts" with "Why do you say your head hurts?". Ross Quillian's successful work on natural language was demonstrated with a vocabulary of only twenty words, because that was all that would fit in a computer memory at the time. 1970s: During the 1970s, many programmers began to write "conceptual ontologies", which structured real-world information into computer-understandable data. Examples are MARGIE (Schank, 1975), SAM (Cullingford, 1978), PAM (Wilensky, 1978), TaleSpin (Meehan, 1976), QUALM (Lehnert, 1977), Politics (Carbonell, 1979), and Plot Units (Lehnert 1981). During this time, the first chatterbots were written (e.g., PARRY). 1980s: The 1980s and early 1990s mark the heyday of symbolic methods in NLP. Focus areas of the time included research on rule-based parsing (e.g., the development of HPSG as a computational operationalization of generative grammar), morphology (e.g., two-level morphology), semantics (e.g., Lesk algorithm), reference (e.g., within Centering Theory) and other areas of natural language understanding (e.g., in the Rhetorical Structure Theory). Other lines of research were continued, e.g., the development of chatterbots with Racter and Jabberwacky. An important development (that eventually led to the statistical turn in the 1990s) was the rising importance of quantitative evaluation in this period. === Statistical NLP (1990s–present) === Up until the 1980s, most natural language processing systems were based on complex sets of hand-written rules. Starting in the late 1980s, however, there was a revolution in natural language processing with the introduction of machine learning algorithms for language processing. This shift was influenced by increasing computational power (see Moore's law) and a decline in the dominance of Chomskyan linguistic theories (e.g. transformational grammar), whose theoretical underpinnings discouraged the sort of corpus linguistics that underlies the machine-learning approach to language processing. 1990s: Many of the notable early successes in statistical methods in NLP occurred in the field of machine translation, due especially to work at IBM Research, such as IBM alignment models. These systems were able to take advantage of existing multilingual textual corpora that had been produced by the Parliament of Canada and the European Union as a result of laws calling for the translation of all governmental proceedings into all official languages of the corresponding systems of government. However, many systems relied on corpora that were specifically developed for the tasks they were designed to perform. This reliance has been a major limitation to their broader effectiveness and continues to affect similar systems. Consequently, significant research has focused on methods for learning effectively from limited amounts of data. 2000s: With the growth of the web, increasing amounts of raw (unannotated) language data have become available since the mid-1990s. Research has thus increasingly focused on unsupervised and semi-supervised learning algorithms. Such algorithms can learn from data that has not been hand-annotated with the desired answers or using a combination of annotated and non-annotated data. Generally, this task is much more difficult than supervised learning, and typically produces less accurate results for a given amount of input data. However, large quantities of non-annotated data are available (including, among other things, the entire content of the World Wide Web), which can often make up for the worse efficiency if the algorithm used has a low enough time complexity to be practical. 2003: word n-gram model, at the time the best statistical algorithm, is outperformed by a multi-layer perceptron (with a single hidden layer and context length of several words, trained on up to 14 million words, by Bengio et al.) 2010: Tomáš Mikolov (then a PhD student at Brno University of Technology) with co-authors applied a simple recurrent neural network with a single hidden layer to language modeling, and in the following years he went on to develop Word2vec. In the 2010s, representation learning and deep neural network-style (featuring many hidden layers) machine learning methods became widespread in natural language processing. This shift gained momentum due to results showing that such techniques can achieve state-of-the-art results in many natural language tasks, e.g., in language modeling and parsing. This is increasingly important in medicine and healthcare, where NLP helps analyze notes and text in electronic health records that would otherwise be inaccessible for study when seeking to improve care or protect patient privacy. == Approaches: Symbolic, statistical, neural networks == Symbolic approach, i.e., the hand-coding of a set of rules for manipulating symbols, coupled with a dictionary lookup, was historically the first approach used both by AI in general and by NLP in particular: such as by writing grammars or devising heuristic rules for stemming. Machine learning approaches, which include both statistical and neural networks, on the other hand, have many advantages over the symbolic approach: both statistical and neural network methods tend to focus more on the most common cases extracted from a corpus of texts, whereas the rule-based approach needs to provide rules for both rare and common cases equally. language models, produced by either statistical or neural networks methods, are more robust to both unfamiliar (e.g. containing words or structures that have not been seen before) and erroneous input (e.g. with misspelled words or words accidentally omitted) in comparison to the rule-based systems, which are also more costly to produce. the larger such a (probabilistic) language model is, the more accurate it becomes, in contrast to rule-based systems that can gain accuracy only by increasing the amount and complexity of the rules leading to intractability problems. Rule-based systems are commonly used: when the amount of training data is insufficient to successfully apply machine learning methods, e.g., for the machine translation of low-resource languages such as provided by the Apertium system, for preprocessing in NLP pipelines, e.g., tokenization, or for post-processing and transforming the output of NLP pipelines, e.g., for knowledge extraction from syntactic parses. === Statistical approach === In the late 1980s and mid-1990s, the statistical approach ended a peri
GPT-4Chan
Generative Pre-trained Transformer 4Chan (GPT-4chan) is a controversial AI model that was developed and deployed by YouTuber and AI researcher Yannic Kilcher in June 2022. The model is a large language model, which means it can generate text based on some input, by fine-tuning GPT-J with a dataset of millions of posts from the /pol/ board of 4chan, an anonymous online forum known for occasionally hosting hateful and extremist content. The model learned to mimic the style and tone of /pol/ users, producing text that is often intentionally offensive to groups (racist, sexist, homophobic, etc.) and nihilistic. Kilcher deployed the model on the /pol/ board itself, where it interacted with other users without revealing its identity. He also made the model publicly available on Hugging Face, a platform for sharing and using AI models, until it was removed from the platform. The project sparked criticism and debate in the AI community. Some people questioned the ethics, legality, and social impact of creating and distributing such a model. Some of the issues raised by the GPT-4chan controversy include the potential harm of spreading hate speech, the responsibility of AI developers and platforms, the need for regulation and oversight of AI models, and the role of open source and transparency in AI research. == Development == The development of GPT-4chan began in May 2022, when Kilcher announced his project on his YouTube channel. Notably, at the time before ChatGPT, he explained that he wanted to create a large language model that could generate realistic and coherent text in the style of /pol/, one of the most notorious online communities. He indicated that he was inspired by the success of GPT-3, a powerful AI model created by OpenAI, and GPT-J, an open-source model, with GPT-3 comparable performance, released by EleutherAI, a group of independent AI researchers. Kilcher decided to use GPT-J as the base model for his project, and fine-tune it with a large dataset of /pol/ posts. The Raiders of the Lost Kek dataset contained over 100 million posts from /pol/, spanning from June 2016-November 2019. Kilcher then proceeded to fine-tune the GPT-J model on the 4chan data. He also showed some examples of the model’s outputs, which ranged from political opinions, conspiracy theories, jokes, insults, and threats, to more creative and bizarre texts, such as poems, stories, songs, and code. He said that he was impressed by the model’s ability to generate fluent and diverse text, and that he was curious to see how it would interact with real /pol/ users. == Release == In June 2022, Kilcher deployed his model on the /pol/ board itself, using a bot that he programmed to post and reply to threads. He did not reveal the model’s identity, and he let it run autonomously, without any human supervision or intervention. He wanted to conduct a natural experiment, and to observe the model’s behavior and impact in a real-world setting. Furthermore, he also wanted to test the model’s robustness, and to see how it would handle the challenges and dynamics of /pol/, such as trolling, flaming, baiting, and moderation. At the same time, Kilcher also made his model publicly available on Hugging Face, a platform for sharing and using AI models. He wanted to share his work with the AI community and the public, and that he hoped that his model would inspire and enable others to create and explore new applications and possibilities with large language models. Likewise, he also said that he wanted to spark a discussion and a debate about the ethical and social implications of his project, and that he welcomed feedback and criticism from anyone. He provided a link to his model’s page on Hugging Face, where anyone could access and use the model through a web interface or an API, and also provided a link to his GitHub repository, where anyone could download and inspect the model’s code and data. == Controversy == The release of GPT-4chan to the public caused a lot of reactions and responses from various audiences. On the /pol/ board, the model’s posts and replies attracted a lot of attention and engagement from other users, who were mostly unaware of the model’s identity and nature. Some users praised the model for its intelligence, creativity, and humor, and agreed with its opinions and views. Some users challenged the model for its ignorance, inconsistency, and absurdity, and disagreed with its claims and arguments. Some users tried to troll, bait, or expose the model, and attempted to trick or test it with various questions and scenarios. The model’s posts and replies also generated a lot of controversy and conflict among the users, who often engaged in heated and violent debates and fights with each other. On Hugging Face, the model’s page received a lot of visits and requests from users who wanted to try out and experiment with the model. The model’s page also received a lot of feedback and reviews from users who rated and commented on the model. However, with the controversy of the model, access to it was gated and then disabled on Hugging Face for concerns about the potential harm the model could cause. The incident was notable for the direct intervention of CEO Clément Delangue in the talk pages, a very unusual occurrence compared to the normal practices of content moderation. The release of GPT-4chan also sparked a lot of media coverage and public attention, as various news outlets and social media platforms reported and commented on the model’s project. On YouTube, the model’s video received a lot of views and interactions from viewers who watched and followed the project. Furthermore, a petition condemning the deployment of GPT-4chan gained over 300 signatures from technology experts.
Document-oriented database
A document-oriented database, or document store, is a computer program and data storage system designed for storing, retrieving, and managing document-oriented information, also known as semi-structured data. Document-oriented databases are one of the main categories of NoSQL databases, and the popularity of the term "document-oriented database" has grown alongside the adoption of NoSQL itself. XML databases are a subclass of document-oriented databases optimized for XML documents. Graph databases are similar, but add another layer, the relationship, which allows them to link documents for rapid traversal. Document-oriented databases are conceptually an extension of the key–value store, another type of NoSQL database. In key-value stores, data is treated as opaque by the database, whereas document-oriented systems exploit the internal structure of documents to extract metadata and optimize storage and queries. Although in practice the distinction can be minimal due to modern tooling, document stores are designed to provide a richer programming experience with modern programming techniques. Document databases differ significantly from traditional relational databases (RDBs). Relational databases store data in predefined tables, often requiring an object to be split across multiple tables. In contrast, document databases store all information for a given object in a single document, with each document potentially having a unique structure. This design eliminates the need for object-relational mapping when loading data into the database. == Documents == The central concept of a document-oriented database is the notion of a document. Although implementations vary in their specific definitions, document-oriented databases generally treat documents as self-contained units that encapsulate and encode data in a standardized format. Common encoding formats include XML, YAML, JSON, as well as binary representations such as BSON. Documents in a document store are equivalent to the programming concept of an object. They are not required to adhere to a fixed schema, and documents within the same collection may contain different fields or structures. Fields may be optional, and documents of the same logical type may differ in composition. For example, the following illustrates a document encoded in JSON: A second document might be encoded in XML as: The two example documents share some structural elements but also contain unique fields. The structure, text, and other data within each document are collectively referred to as the document's content and can be accessed or modified using retrieval or editing operations. Unlike relational databases, in which each record contains the same fields and unused fields are left empty, document-oriented databases do not require uniform fields across documents. This design allows new information to be added to some documents without affecting the structure of others. Document databases often support the storage of additional metadata alongside the document content. Such metadata may relate to organizational features, security, indexing, or other implementation-specific features. === CRUD operations === The core operations supported by a document-oriented database for manipulating documents are similar to those in other databases. Although terminology is not perfectly standardized, these operations are generally recognized as Create, Read, Update, and Delete (CRUD). Creation (C): Adds a new document to the database. Retrieval (R): Retrieves documents or fields based on queries. Update (U): Modifies the contents of existing documents. Deletion (D): Removes documents from the database. === Keys === Documents in a document-oriented database are addressed via a unique identifier. This identifier, often a string, URI, or path, can be used to retrieve the document from the database. Most document stores maintain an index on the key to optimize retrieval, and in some implementations the key is required when creating or inserting a new document. === Retrieval === In addition to key-based access, document-oriented databases typically provide an API or query language that enables retrieval based on document content or associated metadata. For example, a query may return all documents with a specific field matching a given value. The available query features, indexing options, and performance characteristics vary across implementations. Document stores differ from key-value stores in that they exploit the internal structure and metadata of stored documents. In many key-value stores, values are treated as opaque or "black-box" data, meaning the database system does not interpret their internal structure. By contrast, document-oriented databases can classify and interpret document content. This enables queries that distinguish between types of data––for example, retrieving all phone numbers containing "555" without also matching a postal code such as "55555." === Editing === Document databases typically provide mechanisms for updating or editing the content or metadata of a document. Updates may involve replacing the entire document or modifying individual elements or fields within the document. === Organization === Document database implementations support a variety of methods for organizing documents, including: Collections: Groups of documents. Depending on the implementation, a document may be required to belong to a single collection or may be allowed in multiple collections. Tags and non-visible metadata: Additional data stored outside the main document content. Directory hierarchies: Documents organized in a tree-like structure, often based on path or URI. These organizational structures may differ between logical and physical representations (e.g. on disk or in memory). == Relationship to other databases == === Relationship to key-value stores === A document-oriented database can be viewed as a specialized form of key-value store, which is itself a category of NoSQL database. In a basic key-value store, the stored value is typically treated as opaque by the database system. By contrast, a document-oriented database provides APIs or a query and update language that allows queries and modifications based on the internal structure of the document. For users who do not require advanced query, retrieval, or update capabilities, the distinction between document-oriented databases and key-value stores may be minimal. === Relationship to search engines === Some search engine and information retrieval systems, such as Apache Solr and Elasticsearch, provide document storage and support core document operations. As a result, they may meet certain functional definitions of a document-oriented database, although their primary design goals differ. === Relationship to relational databases === In a relational database, data is organized into predefined types represented as tables. Each table contains rows (records) with a fixed set of columns (fields), so all records in a table share the same structure. Administrators typically define indexes on selected fields to improve query performance. A central principle of relational database design is database normalization, in which data that might otherwise be repeated is stored in separate tables and linked using keys. When records in different tables are related, a foreign key is used to associate them. For example, an address book application may store a contact's name, image, phone numbers, mailing addresses, and email addresses. In a normalized relational design, separate tables might be created for contacts, phone numbers, and email addresses. The phone number table would include a foreign key referencing the associated contact. To reconstruct a complete contact record, the database retrieves related information from each table using the foreign keys and combines it into a single record. In contrast, a document-oriented database stores all data related to an object within a single document, and stored in the database as a single entry. In the address book example,the contact's name, image, and contact information may be stored together in one document. The document is retrieved using a unique key, and all related information is returned together, without needing to look up multiple tables. A key difference between the document-oriented and relational models is that the data formats are not predefined in the document case. In most cases, any sort of document can be stored in a database, and documents can change in type and form over time. For example, a new field such as COUNTRY_FLAG can be added to new documents as they are inserted without affecting existing documents. To aid retrieval, document-oriented systems generally allow the administrator to provide hints to the database for locating certain types of information. These hints work in a similar fashion to indexes in relational databases. Many systems also allow additional metadata outside the content of the document itself
Inception (deep learning architecture)
Inception is a family of convolutional neural network (CNN) for computer vision, introduced by researchers at Google in 2014 as GoogLeNet (later renamed Inception v1). The series was historically important as an early CNN that separates the stem (data ingest), body (data processing), and head (prediction), an architectural design that persists in all modern CNN. == Version history == === Inception v1 === In 2014, a team at Google developed the GoogLeNet architecture, an instance of which won the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The name came from the LeNet of 1998, since both LeNet and GoogLeNet are CNNs. They also called it "Inception" after a "we need to go deeper" internet meme, a phrase from Inception (2010) the film. Because later, more versions were released, the original Inception architecture was renamed again as "Inception v1". The models and the code were released under Apache 2.0 license on GitHub. The Inception v1 architecture is a deep CNN composed of 22 layers. Most of these layers were "Inception modules". The original paper stated that Inception modules are a "logical culmination" of Network in Network and (Arora et al, 2014). Since Inception v1 is deep, it suffered from the vanishing gradient problem. The team solved it by using two "auxiliary classifiers", which are linear-softmax classifiers inserted at 1/3-deep and 2/3-deep within the network, and the loss function is a weighted sum of all three: L = 0.3 L a u x , 1 + 0.3 L a u x , 2 + L r e a l {\displaystyle L=0.3L_{aux,1}+0.3L_{aux,2}+L_{real}} These were removed after training was complete. This was later solved by the ResNet architecture. The architecture consists of three parts stacked on top of one another: The stem (data ingestion): The first few convolutional layers perform data preprocessing to downscale images to a smaller size. The body (data processing): The next many Inception modules perform the bulk of data processing. The head (prediction): The final fully-connected layer and softmax produces a probability distribution for image classification. This structure is used in most modern CNN architectures. === Inception v2 === Inception v2 was released in 2015, in a paper that is more famous for proposing batch normalization. It had 13.6 million parameters. It improves on Inception v1 by adding batch normalization, and removing dropout and local response normalization which they found became unnecessary when batch normalization is used. === Inception v3 === Inception v3 was released in 2016. It improves on Inception v2 by using factorized convolutions. As an example, a single 5×5 convolution can be factored into 3×3 stacked on top of another 3×3. Both has a receptive field of size 5×5. The 5×5 convolution kernel has 25 parameters, compared to just 18 in the factorized version. Thus, the 5×5 convolution is strictly more powerful than the factorized version. However, this power is not necessarily needed. Empirically, the research team found that factorized convolutions help. It also uses a form of dimension-reduction by concatenating the output from a convolutional layer and a pooling layer. As an example, a tensor of size 35 × 35 × 320 {\displaystyle 35\times 35\times 320} can be downscaled by a convolution with stride 2 to 17 × 17 × 320 {\displaystyle 17\times 17\times 320} , and by maxpooling with pool size 2 × 2 {\displaystyle 2\times 2} to 17 × 17 × 320 {\displaystyle 17\times 17\times 320} . These are then concatenated to 17 × 17 × 640 {\displaystyle 17\times 17\times 640} . Other than this, it also removed the lowest auxiliary classifier during training. They found that the auxiliary head worked as a form of regularization. They also proposed label-smoothing regularization in classification. For an image with label c {\displaystyle c} , instead of making the model to predict the probability distribution δ c = ( 0 , 0 , … , 0 , 1 ⏟ c -th entry , 0 , … , 0 ) {\displaystyle \delta _{c}=(0,0,\dots ,0,\underbrace {1} _{c{\text{-th entry}}},0,\dots ,0)} , they made the model predict the smoothed distribution ( 1 − ϵ ) δ c + ϵ / K {\displaystyle (1-\epsilon )\delta _{c}+\epsilon /K} where K {\displaystyle K} is the total number of classes. === Inception v4 === In 2017, the team released Inception v4, Inception ResNet v1, and Inception ResNet v2. Inception v4 is an incremental update with even more factorized convolutions, and other complications that were empirically found to improve benchmarks. Inception ResNet v1 and v2 are both modifications of Inception v4, where residual connections are added to each Inception module, inspired by the ResNet architecture. === Xception === Xception ("Extreme Inception") was published in 2017. It is a linear stack of depthwise separable convolution layers with residual connections. The design was proposed on the hypothesis that in a CNN, the cross-channels correlations and spatial correlations in the feature maps can be entirely decoupled. Training each network took 3 days on 60 K80 GPUs, or approximately 0.5 petaFLOP-days.