Consumer relationship systems (CRS) are specialized customer relationship management (CRM) software applications that are used to handle a company's dealings with its customers. Current consumer relationship systems integrate the software with telephone and call recording systems as well as with corporate systems for input and reporting. Customers can provide input from the company's website directly into the CRS. These systems are popular because they can deliver the 'voice of the consumer' that contributes to product quality improvement and that ultimately increases corporate profits. Consumer relationship systems that provide automated support as well as advanced systems may have artificial intelligence (AI) interfaces that can extract and analyse data collected, or handle basic questions and complaints. == History == The first CRS was developed in the 1980s. In 1981 Michael Wilke and Robert Thornton founded Wilke/Thornton, Inc in Columbus, Ohio, to develop new CRS software.
Transderivational search
Transderivational search (often abbreviated to TDS) is a psychological and cybernetics term, meaning when a search is being conducted for a fuzzy match across a broad field. In computing the equivalent function can be performed using content-addressable memory. Unlike usual searches, which look for literal (i.e. exact, logical, or regular expression) matches, a transderivational search is a search for a possible meaning or possible match as part of communication, and without which an incoming communication cannot be made any sense of whatsoever. It is thus an integral part of processing language, and of attaching meaning to communication. In NLP (Neuro-linguistic programming), a transderivational search (Bandler and Grinder, 1976) is essentially the process of searching back through one's stored memories and mental representations to find the personal reference experiences from which a current understanding or mental map has been derived. By the end of 1976, Grinder and Bandler had combined Satir’s and Perls’ language patterns and Erickson’s hypnotic language and use of metaphor with anchoring to create new processes that they called collapsing anchors, trans-derivational search, changing personal history, and reframing. A psychological example of TDS is in Ericksonian hypnotherapy, where vague suggestions are used that the patient must process intensely in order to find their own meanings, thus ensuring that the practitioner does not intrude his own beliefs into the subject's inner world. == TDS in human communication and processing == Because TDS is a compelling, automatic and unconscious state of internal focus and processing (i.e. a type of everyday trance state), and often a state of internal lack of certainty, or openness to finding an answer (since something is being checked out at that moment), it can be utilized or interrupted, in order to create, or deepen, trance. TDS is a fundamental part of human language and cognitive processing. Arguably, every word or utterance a person hears, for example, and everything they see or feel and take note of, results in a very brief trance while TDS is carried out to establish a contextual meaning for it. === Examples === Leading statements: "And those thoughts you had yesterday..." the human mind cannot process hearing this phrase, without at some level searching internally for some thoughts or other that it had yesterday, to make the subject of the sentence. "The many colors that fruit can be" likewise starts the human mind considering even if briefly, different fruit sorted by color. "You did it again, didn't you!" This everyday manipulative use of TDS usually sends the recipient looking internally for some "it" they may have done for which blame is being fairly given. Regardless of whether such a matter can be identified, guilt or anger may result. "There has been pain, hasn't there" the mind of a patient suffering an illness will find it very hard or impossible to hear or answer this sentence without conducting internal searches to verify whether this is true or not, or to find an example if so. "You'd forgotten something [or: some part of your body], hadn't you?" the mind usually checks through the various things, or parts of the body, on hearing this, seeing if each in turn has been forgotten. Textual ambiguity: "Do you remember line dancing on the steps?" Without sufficient context, some statements may trigger TDS in order to resolve inherent ambiguity in the interpretation of a posed question. Do I remember a bygone fad called "line dancing on the steps"? Do I remember personally engaging in dancing in the past? Do I remember my routine practice dancing by focusing on the steps of the dance? Do I tend to forget about dancing when I am standing on steps? "Penny-wise and pound the table dance to the beat of a different drummer". The mixing of cliché and stock phrases may trigger TDS in order to reconcile the discrepancies between expected and actual utterances in sequence. Although TDS is often associated with spoken language, it can be induced in any perceptual system. Thus Milton Erickson's "hypnotic handshake" is a technique that leaves the other person performing TDS in search of meaning to a deliberately ambiguous use of touch.
Artificial intelligence industry in Canada
The artificial intelligence industry in Canada is a rapidly expanding sector. Although Canada held a pioneering role in the early development of artificial intelligence, transforming research excellence into broad commercial adoption has proven challenging. Despite globally recognized scientific achievements and a deep pool of skilled experts, by June 2024, Canada recorded the lowest rate of AI integration among OECD countries, with only 12% of firms implementing AI in their products or services. However, AI adoption has shown significant momentum—doubling from mid-2024 to mid-2025, rising from 6.1% to 12.2%. As of September 2025, Statistics Canada indicated that while about one-third of Canadian businesses had no plans to adopt artificial intelligence in the next year, 14.5% reported intentions to begin using AI for producing goods or delivering services. The primary reasons for not moving forward with AI were lack of relevance, insufficient knowledge, and privacy concerns. According to Public Works Canada (PwC), the pace of AI adoption in Canada is roughly three-quarters of the United States rate, highlighting a notable gap between the two countries in business integration of this technology. British-Canadian computer scientist Geoffrey Hinton stated in 2025 that Canadian companies are adopting artificial intelligence at a slower pace, which may result in the loss of the country's early advantages in the field. At the "All In AI" conference held in Montreal in September 2025, the Minister of Artificial Intelligence and Digital Innovation Evan Solomon, described "Building digital sovereignty" as the most pressing democratic issue of the time. He introduced a 26-person task force focused on updating Canada's AI strategy. In their 2024 report " "Learning Together for Responsible Artificial Intelligence" report, the Innovation, Science, and Economic Development Canada stressed that public awareness, trust, and AI literacy are essential for the responsible adoption and governance of AI in Canada. Montreal workshops in 2021 expanded the OECD's 2019 definition of AI as "the set of computer techniques that enable a machine (e.g., a computer or telephone) to perform tasks that typically require intelligence, such as reasoning or learning. It is also referred to as the automation of intelligent tasks. Scientific developments in AI, such as deep-learning techniques, have made it possible to design access to huge amounts of data and ever-increasing computing power. These new techniques have been rapidly deployed on a large scale in all areas of social life, in transport, education, culture and health." == Federal investments and policy == The 2025 federal budget allocates over $1 billion over the next five years to bolster Canada's artificial intelligence and quantum computing ecosystem. == Industry landscape or research hubs == AlexNet, an influential deep convolutional neural network developed at the University of Toronto by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton, marked a pivotal turning point in modern artificial intelligence. In 2012, it achieved a dramatic reduction in error rates for the ImageNet Large Scale Visual Recognition Challenge (ILSVRC), showcasing the practical power of deep learning and GPU acceleration. The success of AlexNet helped cement Canada’s reputation for AI leadership and inspired rapid adoption of deep learning across the technology sector, with ongoing impact in both academic and commercial domains. In healthcare, AlexNet has been adapted for medical imaging to assist with analyzing radiographs, mammograms, and other scans, including identifying abnormalities and supporting clinical diagnosis. In 2015, the Ottawa-based start-up Advanced Symbolics Inc. (ASI) began developing Polly, an artificial intelligence system designed to analyze and anticipate how target audiences behave—enabling more effective communication strategies and advertising campaigns. Polly was named after its first assignment analyzing the politics of Brexit. The AI gained widespread attention in 2016 for accurately forecasting both the Brexit referendum and the 2016 U.S. presidential election won by Donald Trump. The company states that Polly is used by organizations in diverse sectors—including healthcare, politics, entertainment, and mental health research—to support decision-making based on predictive analytics. Chartwatch, an AI tool developed in Canada, has been shown to reduce unexpected hospital deaths by 26% according to a 2024 study. The system analyzes patient data to detect subtle signs of deterioration, supporting healthcare teams in providing timely interventions. === Notable figures in AI in Canada === Geoffrey Hinton's decades-long work eventually formed the foundation of artificial intelligence, which earned him the Nobel Prize for physics in 2024. Yoshua Bengio, who won the Turing Award in 2018 for his pioneering work in deep learning, founded what would become Mila in 1993. Mila, is currently a collaboration between four Montreal-based academic partners.—the Pan-Canadian Artificial Intelligence Strategy includes Alberta's Amii, Toronto's Vector Institute, and Mila. Fakhreddine Karray's work on operational AI has had tangible impact across several Canadian-relevant sectors, notably intelligent transportation systems, virtual healthcare, and driver safety. === AI in the oil and gas industry === According to a 2020 Ernst & Young report the oil and gas industry in Canada is using AI in automating routine, repetitive, and dangerous tasks with technologies like robotic process automation and machine learning; optimizing production and processing; enhancing transportation logistics; improving equipment operation and monitoring; and enabling preventative maintenance. AI is also deployed for data analysis to improve prediction and decision-making, and is expected to automate up to 50% of job competencies in upstream oil and gas by 2040. Oilsands giant Suncor Energy operates a large fleet of autonomous trucks and has started using AI in its dispatch system at the Mildred Lake mine. As of 2024, AI manages routine tasks such as allocating trucks to dump stations and sending them to refuelling locations. === Indigenous and Inuit Innovation in AI === Indigenous organizations have been working on the creation of new technologies for language revitalization in partnership with National Research Council of Canada since the mid-2010s. In 2025, Inuit researchers and technology partners launched an AI-powered initiative to support the revitalization and preservation of Inuktitut, demonstrating how artificial intelligence can be adapted for Indigenous language and cultural priorities. A 2025 CBC article notes that, while AI can help revitalize Inuktitut, Inuit leaders emphasize concerns about data sovereignty, information ownership, and the need for Indigenous leadership to ensure transparency, privacy, and accountability in AI development. == Regulation == Canada's Artificial Intelligence and Data Act (AIDA) was proposed in November 2022, as part of the Digital Charter Implementation Act (Bill C-27). As well voluntary codes, such as the September 2023 Code of Conduct for Generative AI, and landmark investments in advanced computing infrastructure and the Canadian Artificial Intelligence Safety Institute (CAISI) reflect Canada's commitment to both safety and global competitiveness. == AI infrastructure == Canada has undertaken efforts to expand its AI computing infrastructure at both provincial and federal levels. The federal government's Canadian Sovereign AI Compute Strategy, allocated up to C$2 billion in Budget 2024, aims to enhance computing capacity to support domestic AI industry growth and AI adoption across the economy, with up to C$700 million designated to mobilize private sector investment in new or expanded data centres. Alberta has introduced an AI Data Centres Strategy to position itself as a leading North American destination for data centre investment, targeting C$100 billion worth of AI data centres under development by 2030. One major project under Alberta's strategy is the Wonder Valley AI Data Centre Park near Grande Prairie, which was exempted from provincial environmental impact assessment in April 2026 but still requires permits demonstrating safe construction and operation. According to Statista, as of April 2026, Canada has 287 data centres.
Documentation science
Documentation science is the study of the recording and retrieval of information. It includes methods for storing, retrieving, and sharing of information captured on physical as well as digital documents. This field is closely linked to the fields of library science and information science but has its own theories and practices. The term documentation science was coined by Belgian lawyer and peace activist Paul Otlet. He is considered to be the forefather of information science. He along with Henri La Fontaine laid the foundations of documentation science as a field of study. Professionals in this field are called documentalists. Over the years, documentation science has grown to become a large and important field of study. Evolving from traditional practices like archiving and retrieval to modern theories about the nature of documents, novel methods for organizing digital information, and applications in libraries, research, healthcare, business, and technology and more. This field continues to evolve in the digital age. == Developments in documentation science == 1895: The International Institute of Bibliography (originally Institut International de Bibliographie, IIB) was established on 12 September 1895, in Brussels, Belgium by Paul Otlet and Henri La Fontaine. It aimed to catalog all recorded knowledge using a universal classification system now known as the Universal Decimal Classification (UDC). 1931: International Institute of Bibliography (originally Institut International de Bibliographie, IIB) was renamed The International Institute for Documentation, (Institut International de Documentation, IID). 1934: Paul Otlet envisioned a “radiated library,” a global network of interconnected documents accessible from anywhere via telecommunication. This early idea is now seen as a forerunner of the internet. 1937: American Documentation Institute was founded (1968 nameshift to American Society for Information Science). 1951: Suzanne Briet published Qu'est-ce que la documentation? where she proposed that “a document is evidence in support of a fact,” expanding the definition to include objects such as animals in zoos when they are part of a scientific study. This was a significant theoretical shift in defining documents. 1965-1990: Documentation departments were established, for example, large research libraries, online computer retrieval systems and more. The persons doing the searches were called documentalists. But with the appearance of first CD-ROM databases in the mid-1980s and later the internet in 1990s, these intermediary searches decreased and most such departments closed or merged with other departments. 1996: "Dokvit", Documentation Studies, was established in 1996 at the University of Tromsø in Norway. 2001: The Document Academy was established. It is an international network that celebrates documentation. It was conducted by The Program of Documentation Studies, University of Tromsø, Norway and The School of Information Management and Systems, UC Berkeley. 2003: The first Document Research Conference (DOCAM), a series of conferences made by the Document Academy. DOCAM '03 (2003) was held 13–15 August 2003 at The School of Information Management and Systems (SIMS) at the University of California, Berkeley. 2007: Michael Buckland, Ronald Day, and Birger Hjørland expanded the theoretical foundations of documentation science. They researched and explored documents to be social artifacts, the role of ideology in classification, and how documents influenced knowledge systems. 2010s: The concept of post-documentation or “documentality” began in the 2010s, which focused on how digital traces (e.g., tweets, logs) function as documents without traditional physical form. This led to new thinking in document theory. 2016–present: The Document Academy's DOCAM conferences have continued, offering ongoing developments in the theory and practice of documentation. Themes include affect, memory, activism, and born-digital records. 2017: The journal Information Research published special issues addressing “document theory,” including views on documentation in virtual environments and digital archives. 2020–present: The growth of research data management (RDM) and open science has made documentation practices central to data sharing, metadata standards, and reproducibility in scientific work. == Theoretical foundations == Documentation science has some deep theories that explain what a document is, how people use documents, and how they are organized. These concepts were introduced by scholars who have not only studied libraries, but also philosophy, language, and social sciences. Suzanne Briet described a document as “any material form of evidence” that is made to be used as proof or to share information. An antelope in a zoo, for example, can be a document because it is being studied, classified, and described. Documents are not just things or materials but are also shaped by society. Michael Buckland noted that documents have meaning only when people agree they are useful or valid as information. He explained a document becomes a document when someone decides to use it as evidence. Ronald Day wrote about how documentation is not neutral, it can be influenced by power, ideology, and politics. He claimed that classification systems, like how libraries organize books, are not just technical tools. They also show what kinds of knowledge are seen as more important than others. In recent years, new theories have been introduced, like “documentality” by Maurizio Ferraris. He proposed that a document does not have to be a paper or file, it can also be something digital like a tweet, a database entry, or a log file, as long as it leaves a trace that can be looked at later. This theory helps explain modern digital documents. == Methodologies and practice == Documentation science includes many methods that help people collect, organize, store, and find information. These practices are used in libraries, archives, research labs, companies, and now also in online systems. === Collecting and creating documents === In the past, documentation work included gathering books, articles, reports, and other printed materials. People created records of these materials manually, using catalog cards, indexes, or bibliographies. Paul Otlet’s work with the Universal Bibliographic Repertory is one example. He created millions of card entries to organize knowledge from around the world. Today, documents are not only created by humans. Computers and machines also generate documents, like log files, metadata, and sensor data. These need new tools and methods for collection and management. === Organizing information === Organizing documents has always been a foundational element of documentation science. Methods like classification (dividing things into groups) and indexing (making lists of topics or keywords) help individuals find what they need. A widely used system is the Universal Decimal Classification (UDC) developed by Otlet and La Fontaine. Another is the Library of Congress Classification (LCC) used in the majority of U.S. libraries. Indexing can be performed by humans or by software programs that read the text and add tags to documents. Metadata is also used to describe documents. Metadata is “data about data” like the title, author, date, and subject of a document. Standards like Dublin Core are used in digital libraries to keep metadata consistent. === Retrieval and access === One of the main objectives of documentation is helping users find the right document. This is called information retrieval. In the past, this meant using catalog drawers or printed indexes. Today, people use search engines, databases, and digital libraries. Modern retrieval tools use Boolean logic, ranking algorithms, and sometimes machine learning to show the most useful results first. This is part of what is studied in both documentation science and information retrieval. === Preservation and archiving === Documents require long-term storage. This is called preservation of documents. Printed documents can be damaged by light, pests, or even time on the other hand digital documents can be deemed worthless if formats become outdated or storage facilities fail. Archivists use methods like migration, which includes moving files to new formats, and emulation, which replicates obsolete systems, to preserve materials. These methods and tools are ever changing as new technologies develop. But the main objective of documentation has remained the same, which is to keep information safe, organized, and easy to find. == Documentation in the digital age == With the expansion of the internet, computers, and cloud storage, documents are no longer just books, papers, or reports. They can now be emails, tweets, videos, websites, databases, or even log files created by machines. === Born-digital documents === Many documents today are created directly in digital form. These are called born-digit
Information
Information is an abstract concept that refers to something which has the power to inform. At the most fundamental level, it pertains to the interpretation (perhaps formally) of that which may be sensed, or their abstractions. Any natural process that is not completely random and any observable pattern in any medium can be said to convey some amount of information. Whereas digital signals and other data use discrete signs to convey information, other phenomena and artifacts such as analogue signals, poems, pictures, music or other sounds, and currents convey information in a more continuous form. Information is not knowledge itself, but the meaning that may be derived from a representation through interpretation. The concept of information is relevant to and connected with various concepts, including constraint, communication, control, data, form, education, knowledge, meaning, understanding, mental stimuli, pattern, perception, proposition, representation, and entropy. Information is often processed iteratively: Data available at one step are processed into information to be interpreted and processed at the next step. For example, in written text each symbol or letter conveys information relevant to the word it is part of, each word conveys information relevant to the phrase it is part of, each phrase conveys information relevant to the sentence it is part of, and so on until at the final step information is interpreted and becomes knowledge in a given domain. In a digital signal, bits may be interpreted into the symbols, letters, numbers, or structures that convey the information available at the next level up. The key characteristic of information is that it is subject to interpretation and processing. The derivation of information from a signal or message may be thought of as the resolution of ambiguity or uncertainty that arises during the interpretation of patterns within the signal or message. Information may be structured as data. Redundant data can be compressed up to an optimal size, which is the theoretical limit of compression. The information available through a collection of data may be derived by analysis. For example, a restaurant collects data from every customer order. That information may be analyzed to produce knowledge that is put to use when the business subsequently wants to identify the most popular or least popular dish. Information can be transmitted in time, via data storage, and space, via communication and telecommunication. Information is expressed either as the content of a message or through direct or indirect observation. That which is perceived can be construed as a message in its own right, and in that sense, all information is always conveyed as the content of a message. Information can be encoded into various forms for transmission and interpretation (for example, information may be encoded into a sequence of signs, or transmitted via a signal). It can also be encrypted for safe storage and communication. The uncertainty of an event is measured by its probability of occurrence. Uncertainty is proportional to the negative logarithm of the probability of occurrence. Information theory takes advantage of this by concluding that more uncertain events require more information to resolve their uncertainty. The bit is the standard unit of information. It is 'that which reduces uncertainty by half'. Other units such as the nat may be used. For example, the information encoded in one "fair" coin flip is log2(2/1) = 1 bit, and in two fair coin flips is log2(4/1) = 2 bits. A 2011 Science article estimates that 97% of technologically stored information was already in digital bits in 2007 and that the year 2002 was the beginning of the digital age for information storage (with digital storage capacity bypassing analogue for the first time). == Etymology and history of the concept == The English word "information" comes from Middle French enformacion/informacion/information 'a criminal investigation' and its etymon, Latin informatiō(n) 'conception, teaching, creation'. In English, "information" is an uncountable mass noun. References on "formation or molding of the mind or character, training, instruction, teaching" date from the 14th century in both English (according to Oxford English Dictionary) and other European languages. In the transition from Middle Ages to Modernity the use of the concept of information reflected a fundamental turn in epistemological basis – from "giving a (substantial) form to matter" to "communicating something to someone". Peters (1988, pp. 12–13) concludes: Information was readily deployed in empiricist psychology (though it played a less important role than other words such as impression or idea) because it seemed to describe the mechanics of sensation: objects in the world inform the senses. But sensation is entirely different from "form" – the one is sensual, the other intellectual; the one is subjective, the other objective. My sensation of things is fleeting, elusive, and idiosyncratic. For Hume, especially, sensory experience is a swirl of impressions cut off from any sure link to the real world... In any case, the empiricist problematic was how the mind is informed by sensations of the world. At first informed meant shaped by; later it came to mean received reports from. As its site of action drifted from cosmos to consciousness, the term's sense shifted from unities (Aristotle's forms) to units (of sensation). Information came less and less to refer to internal ordering or formation, since empiricism allowed for no preexisting intellectual forms outside of sensation itself. Instead, information came to refer to the fragmentary, fluctuating, haphazard stuff of sense. Information, like the early modern worldview in general, shifted from a divinely ordered cosmos to a system governed by the motion of corpuscles. Under the tutelage of empiricism, information gradually moved from structure to stuff, from form to substance, from intellectual order to sensory impulses. In the modern era, the most important influence on the concept of information is derived from the Information theory developed by Claude Shannon and others. This theory, however, reflects a fundamental contradiction. Northrup (1993) wrote: Thus, actually two conflicting metaphors are being used: The well-known metaphor of information as a quantity, like water in the water-pipe, is at work, but so is a second metaphor, that of information as a choice, a choice made by :an information provider, and a forced choice made by an :information receiver. Actually, the second metaphor implies that the information sent isn't necessarily equal to the information received, because any choice implies a comparison with a list of possibilities, i.e., a list of possible meanings. Here, meaning is involved, thus spoiling the idea of information as a pure "Ding an sich." Thus, much of the confusion regarding the concept of information seems to be related to the basic confusion of metaphors in Shannon's theory: is information an autonomous quantity, or is information always per SE information to an observer? Actually, I don't think that Shannon himself chose one of the two definitions. Logically speaking, his theory implied information as a subjective phenomenon. But this had so wide-ranging epistemological impacts that Shannon didn't seem to fully realize this logical fact. Consequently, he continued to use metaphors about information as if it were an objective substance. This is the basic, inherent contradiction in Shannon's information theory." (Northrup, 1993, p. 5). In their seminal book The Study of Information: Interdisciplinary Messages, Almach and Mansfield (1983) collected key views on the interdisciplinary controversy in computer science, artificial intelligence, library and information science, linguistics, psychology, and physics, as well as in the social sciences. Almach (1983, p. 660) himself disagrees with the use of the concept of information in the context of signal transmission, the basic senses of information in his view all referring "to telling something or to the something that is being told. Information is addressed to human minds and is received by human minds." All other senses, including its use with regard to nonhuman organisms as well to society as a whole, are, according to Machlup, metaphoric and, as in the case of cybernetics, anthropomorphic. Hjørland (2007) describes the fundamental difference between objective and subjective views of information and argues that the subjective view has been supported by, among others, Bateson, Yovits, Span-Hansen, Brier, Buckland, Goguen, and Hjørland. Hjørland provided the following example: A stone on a field could contain different information for different people (or from one situation to another). It is not possible for information systems to map all the stone's possible information for every individual. Nor is any one mapping the one "true" mapping. But peop
Neural scaling law
In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, and training cost. Some models also exhibit performance gains by scaling inference through increased test-time compute (TTC), extending neural scaling laws beyond training to the deployment phase. == Introduction == In general, a deep learning model can be characterized by four parameters: model size, training dataset size, training cost, and the post-training error rate (e.g., the test set error rate). Each of these variables can be defined as a real number, usually written as N , D , C , L {\displaystyle N,D,C,L} (respectively: parameter count, dataset size, computing cost, and loss). A neural scaling law is a theoretical or empirical statistical law between these parameters. There are also other parameters with other scaling laws. === Size of the model === In most cases, the model's size is simply the number of parameters. However, one complication arises with the use of sparse models, such as mixture-of-expert models. With sparse models, during inference, only a fraction of their parameters are used. In comparison, most other kinds of neural networks, such as transformer models, always use all their parameters during inference. === Size of the training dataset === The size of the training dataset is usually quantified by the number of data points within it. Larger training datasets are typically preferred, as they provide a richer and more diverse source of information from which the model can learn. This can lead to improved generalization performance when the model is applied to new, unseen data. However, increasing the size of the training dataset also increases the computational resources and time required for model training. With the "pretrain, then finetune" method used for most large language models, there are two kinds of training dataset: the pretraining dataset and the finetuning dataset. Their sizes have different effects on model performance. Generally, the finetuning dataset is less than 1% the size of pretraining dataset. In some cases, a small amount of high quality data suffices for finetuning, and more data does not necessarily improve performance. Many scaling laws, due to their inherent diminishing returns nature, value data based on a submodular set function which was shown in a paper on this topic. === Cost of training === Training cost is typically measured in terms of time (how long it takes to train the model) and computational resources (how much processing power and memory are required). It is important to note that the cost of training can be significantly reduced with efficient training algorithms, optimized software libraries, and parallel computing on specialized hardware such as GPUs or TPUs. The cost of training a neural network model is a function of several factors, including model size, training dataset size, the training algorithm complexity, and the computational resources available. In particular, doubling the training dataset size does not necessarily double the cost of training, because one may train the model for several times over the same dataset (each being an "epoch"). === Performance === The performance of a neural network model is evaluated based on its ability to accurately predict the output given some input data. Common metrics for evaluating model performance include: Negative log-likelihood per token (logarithm of perplexity) for language modeling; Accuracy, precision, recall, and F1 score for classification tasks; Mean squared error (MSE) or mean absolute error (MAE) for regression tasks; Elo rating in a competition against other models, such as gameplay or preference by a human judge. Performance can be improved by using more data, larger models, different training algorithms, regularizing the model to prevent overfitting, and early stopping using a validation set. When the performance is a number bounded within the range of [ 0 , 1 ] {\displaystyle [0,1]} , such as accuracy, precision, etc., it often scales as a sigmoid function of cost, as seen in the figures. == Examples == === (Hestness, Narang, et al, 2017) === The 2017 paper is a common reference point for neural scaling laws fitted by statistical analysis on experimental data. Previous works before the 2000s, as cited in the paper, were either theoretical or orders of magnitude smaller in scale. Whereas previous works generally found the scaling exponent to scale like L ∝ D − α {\displaystyle L\propto D^{-\alpha }} , with α ∈ { 0.5 , 1 , 2 } {\displaystyle \alpha \in \{0.5,1,2\}} , the paper found that α ∈ [ 0.07 , 0.35 ] {\displaystyle \alpha \in [0.07,0.35]} . Of the factors they varied, only task can change the exponent α {\displaystyle \alpha } . Changing the architecture optimizers, regularizers, and loss functions, would only change the proportionality factor, not the exponent. For example, for the same task, one architecture might have L = 1000 D − 0.3 {\displaystyle L=1000D^{-0.3}} while another might have L = 500 D − 0.3 {\displaystyle L=500D^{-0.3}} . They also found that for a given architecture, the number of parameters necessary to reach lowest levels of loss, given a fixed dataset size, grows like N ∝ D β {\displaystyle N\propto D^{\beta }} for another exponent β {\displaystyle \beta } . They studied machine translation with LSTM ( α ∼ 0.13 {\displaystyle \alpha \sim 0.13} ), generative language modelling with LSTM ( α ∈ [ 0.06 , 0.09 ] , β ≈ 0.7 {\displaystyle \alpha \in [0.06,0.09],\beta \approx 0.7} ), ImageNet classification with ResNet ( α ∈ [ 0.3 , 0.5 ] , β ≈ 0.6 {\displaystyle \alpha \in [0.3,0.5],\beta \approx 0.6} ), and speech recognition with two hybrid (LSTMs complemented by either CNNs or an attention decoder) architectures ( α ≈ 0.3 {\displaystyle \alpha \approx 0.3} ). === (Henighan, Kaplan, et al, 2020) === A 2020 analysis studied statistical relations between C , N , D , L {\displaystyle C,N,D,L} over a wide range of values and found similar scaling laws, over the range of N ∈ [ 10 3 , 10 9 ] {\displaystyle N\in [10^{3},10^{9}]} , C ∈ [ 10 12 , 10 21 ] {\displaystyle C\in [10^{12},10^{21}]} , and over multiple modalities (text, video, image, text to image, etc.). In particular, the scaling laws it found are (Table 1 of ): For each modality, they fixed one of the two C , N {\displaystyle C,N} , and varying the other one ( D {\displaystyle D} is varied along using D = C / 6 N {\displaystyle D=C/6N} ), the achievable test loss satisfies L = L 0 + ( x 0 x ) α {\displaystyle L=L_{0}+\left({\frac {x_{0}}{x}}\right)^{\alpha }} where x {\displaystyle x} is the varied variable, and L 0 , x 0 , α {\displaystyle L_{0},x_{0},\alpha } are parameters to be found by statistical fitting. The parameter α {\displaystyle \alpha } is the most important one. When N {\displaystyle N} is the varied variable, α {\displaystyle \alpha } ranges from 0.037 {\displaystyle 0.037} to 0.24 {\displaystyle 0.24} depending on the model modality. This corresponds to the α = 0.34 {\displaystyle \alpha =0.34} from the Chinchilla scaling paper. When C {\displaystyle C} is the varied variable, α {\displaystyle \alpha } ranges from 0.048 {\displaystyle 0.048} to 0.19 {\displaystyle 0.19} depending on the model modality. This corresponds to the β = 0.28 {\displaystyle \beta =0.28} from the Chinchilla scaling paper. Given fixed computing budget, optimal model parameter count is consistently around N o p t ( C ) = ( C 5 × 10 − 12 petaFLOP-day ) 0.7 = 9.0 × 10 − 7 C 0.7 {\displaystyle N_{opt}(C)=\left({\frac {C}{5\times 10^{-12}{\text{petaFLOP-day}}}}\right)^{0.7}=9.0\times 10^{-7}C^{0.7}} The parameter 9.0 × 10 − 7 {\displaystyle 9.0\times 10^{-7}} varies by a factor of up to 10 for different modalities. The exponent parameter 0.7 {\displaystyle 0.7} varies from 0.64 {\displaystyle 0.64} to 0.75 {\displaystyle 0.75} for different modalities. This exponent corresponds to the ≈ 0.5 {\displaystyle \approx 0.5} from the Chinchilla scaling paper. It's "strongly suggested" (but not statistically checked) that D o p t ( C ) ∝ N o p t ( C ) 0.4 ∝ C 0.28 {\displaystyle D_{opt}(C)\propto N_{opt}(C)^{0.4}\propto C^{0.28}} . This exponent corresponds to the ≈ 0.5 {\displaystyle \approx 0.5} from the Chinchilla scaling paper. The scaling law of L = L 0 + ( C 0 / C ) 0.048 {\displaystyle L=L_{0}+(C_{0}/C)^{0.048}} was confirmed during the training of GPT-3 (Figure 3.1 ). === Chinchilla scaling (Hoffmann, et al, 2022) === One particular scaling law ("Chinchilla scaling") states that, for a large language model (LLM) autoregressively trained for one epoch, with a cosine learning rate schedule, we have: { C = C 0 N D L = A N α + B D β + L 0 {\displaystyle {\begin{cases}C=C_{0}ND\\L={\frac {A}{N^{\alpha }}}+{\frac {B}{D^{\beta }}}+L_{0}\end{cases}}} where the variables are C {\displaystyle C} is the cost o
Physical schema
A physical data model (or database design) is a representation of a data design as implemented, or intended to be implemented, in a database management system. In the lifecycle of a project it typically derives from a logical data model, though it may be reverse-engineered from a given database implementation. A complete physical data model will include all the database artifacts required to create relationships between tables or to achieve performance goals, such as indexes, constraint definitions, linking tables, partitioned tables or clusters. Analysts can usually use a physical data model to calculate storage estimates; it may include specific storage allocation details for a given database system. As of 2012 seven main databases dominate the commercial marketplace: Informix, Oracle, Postgres, SQL Server, Sybase, IBM Db2 and MySQL. Other RDBMS systems tend either to be legacy databases or used within academia such as universities or further education colleges. Physical data models for each implementation would differ significantly, not least due to underlying operating-system requirements that may sit underneath them. For example: SQL Server runs only on Microsoft Windows operating-systems (Starting with SQL Server 2017, SQL Server runs on Linux. It's the same SQL Server database engine, with many similar features and services regardless of your operating system), while Oracle and MySQL can run on Solaris, Linux and other UNIX-based operating-systems as well as on Windows. This means that the disk requirements, security requirements and many other aspects of a physical data model will be influenced by the RDBMS that a database administrator (or an organization) chooses to use. == Physical schema == Physical schema is a term used in data management to describe how data is to be represented and stored (files, indices, etc.) in secondary storage using a particular database management system (DBMS) (e.g., Oracle RDBMS, Sybase SQL Server, etc.). In the ANSI/SPARC Architecture three schema approach, the internal schema is the view of data that involved data management technology. This is as opposed to an external schema that reflects an individual's view of the data, or the conceptual schema that is the integration of a set of external schemas. The logical schema was the way data were represented to conform to the constraints of a particular approach to database management. At that time the choices were hierarchical and network. Describing the logical schema, however, still did not describe how physically data would be stored on disk drives. That is the domain of the physical schema. Now logical schemas describe data in terms of relational tables and columns, object-oriented classes, and XML tags. A single set of tables, for example, can be implemented in numerous ways, up to and including an architecture where table rows are maintained on computers in different countries.