Confidential computing

Confidential computing is a security and privacy-enhancing computational technique focused on protecting data in use. Confidential computing can be used in conjunction with storage and network encryption, which protect data at rest and data in transit respectively. It is designed to address software, protocol, cryptographic, and basic physical and supply-chain attacks, although some critics have demonstrated architectural and side-channel attacks effective against the technology. The technology protects data in use by performing computations in a hardware-based trusted execution environment (TEE). Confidential data is released to the TEE only once it is assessed to be trustworthy. Different types of confidential computing define the level of data isolation used, whether virtual machine, application, or function, and the technology can be deployed in on-premise data centers, edge locations, or the public cloud. It is often compared with other privacy-enhancing computational techniques such as fully homomorphic encryption, secure multi-party computation, and Trusted Computing. Confidential computing is promoted by the Confidential Computing Consortium (CCC) industry group, whose membership includes major providers of the technology. == Properties == Trusted execution environments (TEEs) "prevent unauthorized access or modification of applications and data while they are in use, thereby increasing the security level of organizations that manage sensitive and regulated data". Trusted execution environments can be instantiated on a computer's processing components such as a central processing unit (CPU) or a graphics processing unit (GPU). In their various implementations, TEEs can provide different levels of isolation including virtual machine, individual application, or compute functions. Typically, data in use in a computer's compute components and memory exists in a decrypted state and can be vulnerable to examination or tampering by unauthorized software or administrators. According to the CCC, confidential computing protects data in use through a minimum of three properties: Data confidentiality: "Unauthorized entities cannot view data while it is in use within the TEE". Data integrity: "Unauthorized entities cannot add, remove, or alter data while it is in use within the TEE". Code integrity: "Unauthorized entities cannot add, remove, or alter code executing in the TEE". In addition to trusted execution environments, remote cryptographic attestation is an essential part of confidential computing. The attestation process assesses the trustworthiness of a system and helps ensure that confidential data is released to a TEE only after it presents verifiable evidence that it is genuine and operating with an acceptable security posture. It allows the verifying party to assess the trustworthiness of a confidential computing environment through an "authentic, accurate, and timely report about the software and data state" of that environment. "Hardware-based attestation schemes rely on a trusted hardware component and associated firmware to execute attestation routines in a secure environment". Without attestation, a compromised system could deceive others into trusting it, claim it is running certain software in a TEE, and potentially compromise the confidentiality or integrity of the data being processed or the integrity of the trusted code. == Technical approaches == Technical approaches to confidential computing may vary in which software, infrastructure and administrator elements are allowed to access confidential data. The "trust boundary," which circumscribes a trusted computing base (TCB), defines which elements have the potential to access confidential data, whether they are acting benignly or maliciously. Confidential computing implementations enforce the defined trust boundary at a specific level of data isolation. The three main types of confidential computing are: Virtual machine isolation Application isolation, also known as process isolation Function isolation, also known as library isolation Virtual machine isolation removes the elements controlled by the computer infrastructure or cloud provider, but allows potential data access by elements inside a virtual machine running on the infrastructure. Application or process isolation permits data access only by authorized software applications or processes. Function or library isolation is designed to permit data access only by authorized subroutines or modules within a larger application, blocking access by any other system element, including unauthorized code in the larger application. == Threat model == As confidential computing is concerned with the protection of data in use, only certain threat models can be addressed by this technique. Other types of attacks are better addressed by other privacy-enhancing technologies. === In scope === The following threat vectors are generally considered in scope for confidential computing: Software attacks: including attacks on the host’s software and firmware. This may include the operating system, hypervisor, BIOS, other software and workloads. Protocol attacks: including "attacks on protocols associated with attestation as well as workload and data transport". This includes vulnerabilities in the "provisioning or placement of the workload" or data that could cause a compromise. Cryptographic attacks: including "vulnerabilities found in ciphers and algorithms due to a number of factors, including mathematical breakthroughs, availability of computing power and new computing approaches such as quantum computing". The CCC notes several caveats in this threat vector, including relative difficulty of upgrading cryptographic algorithms in hardware and recommendations that software and firmware be kept up-to-date. A multi-faceted, defense-in-depth strategy is recommended as a best practice. Basic physical attacks: including cold boot attacks, bus and cache snooping and plugging attack devices into an existing port, such as a PCI Express slot or USB port. Basic upstream supply-chain attacks: including attacks that would compromise TEEs through changes such as added debugging ports. The degree and mechanism of protection against these threats varies with specific confidential computing implementations. === Out of scope === Threats generally defined as out of scope for confidential computing include: Sophisticated physical attacks: including physical attacks that "require long-term and/or invasive access to hardware" such as chip scraping techniques and electron microscope probes. Upstream hardware supply-chain attacks: including attacks on the CPU manufacturing process, CPU supply chain in key injection/generation during manufacture. Attacks on components of a host system that are not directly providing the capabilities of the trusted execution environment are also generally out-of-scope. Availability attacks: confidential computing is designed to protect the confidentiality and integrity of protected data and code. It does not address availability attacks such as Denial of Service or Distributed Denial of Service attacks. == Use cases == Confidential computing can be deployed in the public cloud, on-premise data centers, or distributed "edge" locations, including network nodes, branch offices, industrial systems and others. === Data privacy and security === Confidential computing protects the confidentiality and integrity of data and code from the infrastructure provider, unauthorized or malicious software and system administrators, and other cloud tenants, which may be a concern for organizations seeking control over sensitive or regulated data. The additional security capabilities offered by confidential computing can help accelerate the transition of more sensitive workloads to the cloud or edge locations. === Multi-party analytics === Confidential computing can enable multiple parties to engage in joint analysis using confidential or regulated data inside a TEE while preserving privacy and regulatory compliance. In this case, all parties benefit from the shared analysis, but no party's sensitive data or confidential code is exposed to the other parties or system host. Examples include multiple healthcare organizations contributing data to medical research, or multiple banks collaborating to identify financial fraud or money laundering. Oxford University researchers proposed the alternative paradigm called "Confidential Remote Computing" (CRC), which supports confidential operations in Trusted Execution Environments across endpoint computers considering multiple stakeholders as mutually distrustful data, algorithm and hardware providers. === Confidential generative AI === Confidential computing technologies can be applied to various stages of a generative AI deployments to help increase data or model privacy, security, and regulatory compliance. TEEs and remote attestation can protect the integrity of data during AI model training, keep

Medical data breach

Medical data, including patients' identity information, health status, disease diagnosis and treatment, and biogenetic information, not only involve patients' privacy but also have a special sensitivity and important value, which may bring physical and mental distress and property loss to patients and even negatively affect social stability and national security once leaked. However, the development and application of medical AI must rely on a large amount of medical data for algorithm training, and the larger and more diverse the amount of data, the more accurate the results of its analysis and prediction will be. However, the application of big data technologies such as data collection, analysis and processing, cloud storage, and information sharing has increased the risk of data leakage. In the United States, the rate of such breaches has increased over time, with 176 million records breached by the end of 2017. By 2024, the U.S. Department of Health and Human Services reported 725 large healthcare data breaches affecting approximately 275 million individual records in a single year, marking a significant escalation in both the frequency and scale of incidents. == Black market for health data == In February 2015 an NPR report claimed that organized crime networks had ways of selling health data in the black market. In 2015 a Beazley employee estimated that medical records could sell on the black market for US$40-50. == How data is lost == Theft, data loss, hacking, and unauthorized account access are ways in which medical data breaches happen. Among reported breaches of medical information in the United States networked information systems accounted for the largest number of records breached. There are many data breaches happening in the US health care system, among business associates of the health care providers that continuously gain access to patients' data. == List of data breaches == In February 2024, a ransomware attack on Change Healthcare, a subsidiary of UnitedHealth Group, compromised the protected health information of approximately 100 million individuals, making it the largest healthcare data breach in United States history. The attack disrupted claims processing for healthcare providers nationwide for several weeks. In May 2024, MediSecure suffered a cyberattack involving ransomware in Australia. In May 2021, the Health Service Executive in the Republic of Ireland was the victim of a cyberattack involving ransomware, in the Health Service Executive cyberattack, with admission records and test results present in a sample of the data reviewed by the Financial Times. In October 2018, the Centers for Medicare and Medicaid Services in the US reported that around 75,000 individual records had been affected by a data breach that took place through the ACA Agent and Broker Portal. In 2018, Social Indicators Research published the scientific evidence of 173,398,820 (over 173 million) individuals affected in USA from October 2008 (when the data were collected) to September 2017 (when the statistical analysis took place). In 2015, Anthem Inc. lost data for 37 million people in the Anthem medical data breach In 2014 4.5 million people using Complete Health Systems had their data stolen In 2013-14 1 million people using Montana Department of Public Health and Human Services had their data stolen In 2013 4 million people using Advocate Health and Hospitals Corporation had their data stolen In 2011 4.9 million users of Tricare services had their data stolen due to an employee error by Science Applications International Corporation In 2011 1.9 million people using Health Net had their data stolen In 2011 1 million people using Nemours Foundation had their data stolen In 2010 6800 people using New York-Presbyterian Hospital and Columbia University Medical Center had their data breached. In response, those organizations agreed to pay the United States Department of Health and Human Services a US$4.8 million dollar fine. In 2009 1 million people using BlueCross BlueShield of Tennessee had their data stolen == Regulation == In the United States, the Health Insurance Portability and Accountability Act and Health Information Technology for Economic and Clinical Health Act require companies to report data breaches to affected individuals and the federal government. Under the HIPAA Breach Notification Rule, covered entities must notify affected individuals without unreasonable delay and no later than 60 days after discovering a breach of unsecured protected health information. Breaches affecting 500 or more individuals must also be reported to the HHS Secretary and to prominent media outlets serving the affected state or jurisdiction within the same timeframe; HHS publicly lists these larger breaches on its breach portal, commonly known as the "wall of shame." Breaches affecting fewer than 500 individuals are reported to HHS annually, no later than 60 days after the end of the calendar year in which they were discovered. Health Information Privacy Health Insurance Portability and Accountability Act of 1996 (HIPAA). - 45 CFR Parts 160 and 164, Standards for Privacy of Individually Identifiable Health Information and Security Standards for the Protection of Electronic Protected Health Information. HIPAA includes provisions designed to save health care businesses money by encouraging electronic transactions, as well as regulations to protect the security and confidentiality of patient information. The Privacy Rule became effective April 14, 2001, and most covered entities (health plans, health care clearinghouses, and health care providers that conduct certain financial and administrative transactions electronically) had until April 2003 to comply. This security provision became effective April 21, 2003. The Health Insurance Portability and Accountability Act (HIPAA) is the baseline set of federal regulations governing medical information. It does three things: i. i. i.Establish a structure for how personal health information is disclosed and establish the rights of individuals with respect to health information; ii.Specify security standards for the retention and transmission of electronic patient information; iii.Need a common format and data structure for the electronic exchange of health information. California-Specific Laws California’s medical privacy laws, primarily the Confidentiality of Medical Information Act (CMIA), the data breach sections of the Civil Code, and sections of the Health and Safety Code, provide HIPAA-like protections, although the terminology is different. HIPAA establishes a federal "minimum standard" that applies where there are gaps in California law, and HIPAA also specifies that stricter state laws will override or supersede HIPAA. California's health care privacy laws apply to providers who provide personal health records (PHR), while HIPAA only applies when the provider providing the PHR is a business associate of a covered entity. Federal law does not grant individuals the right to file a lawsuit in the event of a data breach (only the Attorney General can file a lawsuit), but California law does. This means that California law sets a higher standard for medical privacy, and that individuals in California enjoy stronger legal protections and more ways to hold entities that violate their medical privacy accountable. In the UK, the legal framework for how patient data is cared for and processed is the Data Protection Act 2018 (DPA), which incorporates the EU General Data Protection Regulation (GDPR) into law, and the common law duty of confidentiality (CLDC). The data protection legislation requires that the collection and processing of personal data be fair, lawful and transparent. This means that the collection and processing of data as defined by data protection legislation must always have a valid lawful basis and must also meet the requirements of the CLDC. In the China, Article 18 of the "National Health Care Big Data Standards, Security and Services Management Measures (for Trial Implementation)" (National Health Planning and Development (2018) No. 23) promulgated by the National Health Care Commission in 2018 states, "The responsible unit shall adopt measures such as data classification, important data backup, and encryption authentication to guarantee the security of health care big data." However, the scope and definition of important data are not covered. Although the "Information Security Technology-Healthcare Data Security Guide" (the "Guide") issued by the National Standardization Committee also proposes that important data should be evaluated and approved in accordance with the regulations, there is likewise no definition of the connotation and definition of important data.

Statistical relational learning

Statistical relational learning (SRL) is a subdiscipline of artificial intelligence and machine learning that is concerned with domain models that exhibit both uncertainty (which can be dealt with using statistical methods) and complex, relational structure. Typically, the knowledge representation formalisms developed in SRL use (a subset of) first-order logic to describe relational properties of a domain in a general manner (universal quantification) and draw upon probabilistic graphical models (such as Bayesian networks or Markov networks) to model the uncertainty; some also build upon the methods of inductive logic programming. Significant contributions to the field have been made since the late 1990s. As is evident from the characterization above, the field is not strictly limited to learning aspects; it is equally concerned with reasoning (specifically probabilistic inference) and knowledge representation. Therefore, alternative terms that reflect the main foci of the field include statistical relational learning and reasoning (emphasizing the importance of reasoning) and first-order probabilistic languages (emphasizing the key properties of the languages with which models are represented). Another term that is sometimes used in the literature is relational machine learning (RML). == Canonical tasks == A number of canonical tasks are associated with statistical relational learning, the most common ones being. collective classification, i.e. the (simultaneous) prediction of the class of several objects given objects' attributes and their relations link prediction, i.e. predicting whether or not two or more objects are related link-based clustering, i.e. the grouping of similar objects, where similarity is determined according to the links of an object, and the related task of collaborative filtering, i.e. the filtering for information that is relevant to an entity (where a piece of information is considered relevant to an entity if it is known to be relevant to a similar entity) social network modelling object identification/entity resolution/record linkage, i.e. the identification of equivalent entries in two or more separate databases/datasets == Representation formalisms == One of the fundamental design goals of the representation formalisms developed in SRL is to abstract away from concrete entities and to represent instead general principles that are intended to be universally applicable. Since there are countless ways in which such principles can be represented, many representation formalisms have been proposed in recent years. In the following, some of the more common ones are listed in alphabetical order: Bayesian logic program BLOG model Markov logic networks Multi-entity Bayesian network Probabilistic logic programs Probabilistic relational model – a Probabilistic Relational Model (PRM) is the counterpart of a Bayesian network in statistical relational learning. Probabilistic soft logic Recursive random field Relational Bayesian network Relational dependency network Relational Markov network Relational Kalman filtering

Spatial embedding

Spatial embedding is one of feature learning techniques used in spatial analysis where points, lines, polygons or other spatial data types. representing geographic locations are mapped to vectors of real numbers. Conceptually it involves a mathematical embedding from a space with many dimensions per geographic object to a continuous vector space with a much lower dimension. Such embedding methods allow complex spatial data to be used in neural networks and have been shown to improve performance in spatial analysis tasks == Embedded data types == Geographic data can take many forms: text, images, graphs, trajectories, polygons. Depending on the task, there may be a need to combine multimodal data from different sources. The next section describes examples of different types of data and their uses. === Text === Geolocated posts on social media can be used to acquire a library of documents bound to a given place that can be later transformed to embedded vectors using word embedding techniques. === Image === Satellites and aircraft collect digital spatial data acquired from remotely sensed images which can be used in machine learning. They are sometimes hard to analyse using basic image analysis methods and convolutional neural networks can be used to acquire an embedding of images bound to a given geographical object or a region. === Point === A single point of interest (POI) can be assigned multiple features that can be used in machine learning. These could be demographic, transportation, meteorological, or economic data, for example. When embedding single points, it is common to consider the entire set of available points as nodes in a graph. === Line / multiline === Among other things, motion trajectories are represented as lines (multilines). Individual trajectories are embedded taking into account travel time, distances and also features of points visited along the way. Embedding of trajectories allows to improve performance of such tasks as clustering and also categorization. === Polygon === The geographic areas analyzed in machine learning are defined by both administrative boundaries and top-down division into grids of regular shapes such as rectangles, for example. Both types are represented as polygons and, like points, can be assigned different demographic, transportation, or economic features. A polygon can also have features related to the size of the area or shape it represents. === Graph === An example domain where graph representation is used is the street layout in a city, where vertices can be intersections and edges can be roads. The vertices can also be destination points like public transport stops or important points in the city, and the edges represent the flow between them. Embedding graphs or single vertices allows to improve accuracy of analysis methods in which the treated geographical domain can be represented as a network. == Usage == POI recommendation - generating personalized point of interest recommendations based on user preferences. Next/future location prediction - prediction of the next location a person will go to based on their historical trajectory. Zone functions classification - based on different mobility of people or POI distribution a function of a given area in a city can be predicted. Crime prediction - estimation of crime rate in different regions of a city. Local event detection - studying spatio-temporal changes in embeddings can provide valuable information in detection of local event occurring in specific location. Regional mobility popularity prediction - analysis of mobility can show patterns in popularity of different regions in a city. Shape matching - finding a similar shape of given polygon, for example finding building with the same shape as input building. Travel time estimation - predicting estimated travel time given current traffic conditions and special occurring events. Time estimation for on-demand food delivery - estimation of delivery time when placing an order through the website. == Temporal aspect == Some of the data analyzed has a timestamp associated with it. In some cases of data analysis this information is omitted and in others it is used to divide the set into groups. The most common division is the separation of weekdays from weekends or division into hours of the day. This is particularly important in the analysis of mobility data, because the characteristics of mobility during the week and at different times of the day are very different from each other. Another area in which time division into, for example, individual months can be used is in the analysis of tourism of a given region. In order to take such a split into account, embedding methods treat the time stamp specifically or separate versions of the model are developed for different subgroups of the analyzed set.

POP-11

POP-11 is a reflective, incrementally compiled programming language with many of the features of an interpreted language. It is the core language of the Poplog programming environment developed originally by the University of Sussex, and recently in the School of Computer Science at the University of Birmingham, which hosts the main Poplog website. POP-11 is an evolution of the language POP-2, developed in Edinburgh University, and features an open stack model (like Forth, among others). It is mainly procedural, but supports declarative language constructs, including a pattern matcher, and is mostly used for research and teaching in artificial intelligence, although it has features sufficient for many other classes of problems. It is often used to introduce symbolic programming techniques to programmers of more conventional languages like Pascal, who find POP syntax more familiar than that of Lisp. One of POP-11's features is that it supports first-class functions. POP-11 is the core language of the Poplog system. The availability of the compiler and compiler subroutines at run-time (a requirement for incremental compiling) gives it the ability to support a far wider range of extensions (including run-time extensions, such as adding new data-types) than would be possible using only a macro facility. This made it possible for (optional) incremental compilers to be added for Prolog, Common Lisp and Standard ML, which could be added as required to support either mixed language development or development in the second language without using any POP-11 constructs. This made it possible for Poplog to be used by teachers, researchers, and developers who were interested in only one of the languages. The most successful product developed in POP-11 was the Clementine data mining system, developed by ISL. After SPSS bought ISL, they renamed Clementine to SPSS Modeler and decided to port it to C++ and Java, and eventually succeeded with great effort, and perhaps some loss of the flexibility provided by the use of an AI language. POP-11 was for a time available only as part of an expensive commercial package (Poplog), but since about 1999 it has been freely available as part of the open-source software version of Poplog, including various added packages and teaching libraries. An online version of ELIZA using POP-11 is available at Birmingham. At the University of Sussex, David Young used POP-11 in combination with C and Fortran to develop a suite of teaching and interactive development tools for image processing and vision, and has made them available in the Popvision extension to Poplog. == Simple code examples == Here is an example of a simple POP-11 program: define Double(Source) -> Result; Source2 -> Result; enddefine; Double(123) => That prints out: 246 This one includes some list processing: define RemoveElementsMatching(Element, Source) -> Result; lvars Index; [[% for Index in Source do unless Index = Element or Index matches Element then Index; endunless; endfor; %]] -> Result; enddefine; RemoveElementsMatching("the", [[the cat sat on the mat]]) => ;;; outputs [[cat sat on mat]] RemoveElementsMatching("the", [[the cat] [sat on] the mat]) => ;;; outputs [[the cat] [sat on] mat] RemoveElementsMatching([[= cat]], [[the cat]] is a [[big cat]]) => ;;; outputs [[is a]] Examples using the POP-11 pattern matcher, which makes it relatively easy for students to learn to develop sophisticated list-processing programs without having to treat patterns as tree structures accessed by 'head' and 'tail' functions (CAR and CDR in Lisp), can be found in the online introductory tutorial. The matcher is at the heart of the SimAgent (sim_agent) toolkit. Some of the powerful features of the toolkit, such as linking pattern variables to inline code variables, would have been very difficult to implement without the incremental compiler facilities.

Meta-Labeling

Meta-labeling, also known as corrective AI, is a machine learning (ML) technique utilized in quantitative finance to enhance the performance of investment and trading strategies, developed in 2017 by Marcos López de Prado at Guggenheim Partners and Cornell University. The core idea is to separate the decision of trade direction (side) from the decision of trade sizing, addressing the inefficiencies of simultaneously learning both side and size predictions. The side decision involves forecasting market movements (long, short, neutral), while the size decision focuses on risk management and profitability. It serves as a secondary decision-making layer that evaluates the signals generated by a primary predictive model. By assessing the confidence and likely profitability of those signals, meta-labeling allows investors and algorithms to dynamically size positions and suppress false positives. == Motivation == Meta-labeling is designed to improve precision without sacrificing recall. As noted by López de Prado, attempting to model both the direction and the magnitude of a trade using a single algorithm can result in poor generalization. By separating these tasks, meta-labeling enables greater flexibility and robustness: Enhances control over capital allocation. Reduces overfitting by limiting model complexity. Allows the use of interpretability tools and tailored thresholds to manage risk. Enables dynamic trade suppression in unfavorable regimes. == Applications == Meta-labeling has been applied in a variety of financial ML contexts, including: Algorithmic trading: Filtering and sizing trades to reduce false positives. Portfolio optimization: Scaling exposure across multiple signals with differing confidence levels. Risk management: Dynamically disabling strategies in adverse market conditions. Model validation: Interpreting when and why a model may be underperforming due to regime shifts. == General architecture == Meta-labeling decouples two core components of systematic trading strategies: directional prediction and position sizing. The process involves training a primary model to generate trade signals (e.g., buy, sell, or hold) and then training a secondary model to determine whether each signal is likely to lead to a profitable trade. The second model outputs a probability that is interpreted as the confidence in the forecast, which can be used to adjust the position size or to filter out unreliable trades. Meta-labeling is typically implemented as a three-stage process: Primary model (M1): Predicts the direction or label of a financial outcome using features such as market prices, returns, or volatility indicators. A typical output is directional, e.g., Y ∈ {−1,0,1}, representing short, neutral, or long positions. Secondary model (M2): A binary classifier trained to predict whether the primary model's prediction will be profitable. The target variable is a binary meta-label F ∈ { 0 , 1 } {\displaystyle F\in \{0,1\}} . Inputs can include features used in the primary model, performance diagnostics, or market regime data. Position sizing algorithm (M3): Translates the output probability of the secondary model into a position size. Higher confidence scores result in larger allocations, while lower confidence leads to reduced or zero exposure. === Stage 1: Forecasting side === Primary model architecture Figure 1 Figure 1 presents the architecture of a primary model. It focuses on forecasting the side of the trade. Following the example, this model (M1) takes in input data – such as open-high-low-close data and determines the side of the position to take: a negative number is a short position, and positive number is a long position, the range is set between −1 and 1 (the closer it is to −1 or 1, the stronger the models conviction is). When training the model, the labels are −1 and 1, based on the direction of forward returns for some predefined investment horizon. The researcher may decide to apply a recall check (τ: "Tau") by setting a minimum threshold that the initial output needs to be to qualify of a short or long position (if the threshold is not met, no side forecast is predicted, leading to closing of any open positions), this leads to the primary model output which is one of three possible side forecasts: −1, 0, or 1. The primary model also generates evaluation data which can be used by the secondary model, to improve performance of size forecasts. Some examples of evaluation data include rolling accuracy, F1, recall, precision, and AUC scores. === Stage 2: Filtering out false positives === General meta-labeling architecture Figure 2 Next comes the phase of filtering out false positives, by applying a secondary machine learning model (M2), which is a binary classifier trained to determine if the trade will be profitable or not. The model takes as input four general groupings of data: General input data which is predictive of a false positive. For example the last 30 days rolling volatility of the underlying asset. Evaluation data. Market state and regime data, one may find that macro economic data or clustering the market into regimes may help as specific trading strategies are known to perform better in particular regimes. Example: momentum based strategies perform best in periods with low volatility and strong directional moves. Primary models initial input which is a value between −1 and 1. This highlights the strength of the primary models conviction. The output of the model is a value between −1 and 1 (if using a Tanh function) which will indicate the strength of the conviction that a short or long position is profitable, or it could simply be between 0 and 1 (using a sigmoid function) if one only wanted to know if it made money or not. This output allows filtering out trades that are likely to lead to losses. One could stop at this point or use the outputs of the secondary model as inputs to a position sizing algorithm (M3) which could further enhance strategy performance metrics by translating the output probability of the secondary model into a position size. Higher confidence scores result in larger allocations, while lower confidence leads to reduced or zero exposure. === Stage 3: Optimizing position sizes === ==== Position sizing methods (M3) ==== Various algorithms have been proposed for transforming predicted probabilities into trade sizes: All-or-nothing: Allocate 100% of capital if the probability exceeds a predefined threshold (e.g., 0.5); otherwise, do not trade. Model confidence: Use the probability score directly as the fraction of capital allocated. Linear scaling: Rescale the model's probabilities using min-max normalization based on the training data. Normal CDF (NCDF): Use a normal cumulative distribution function applied to a z-statistic derived from the predicted probability. Empirical CDF (ECDF): Rank probabilities based on their percentile in the training data to ensure relative allocation. Sigmoid Optimal Position Sizing (SOPS): Applies a smooth non-linear sigmoid transformation optimized to maximize risk-adjusted returns (Sharpe ratio). ==== Model calibration ==== Each machine learning algorithm used in meta-labeling tends to produce outputs with different characteristic distributions; for example, some are approximately normally distributed, whereas others exhibit a pronounced U-shape, concentrating probabilities near the extremes. Due to these varying distributions, simply summing the outputs of different models can inadvertently lead to uneven weighting of signals, biasing trade decisions. To address this, model calibration techniques are essential to adjust the predicted probabilities towards frequentist probabilities, ensuring that model outputs reflect true likelihoods more accurately. Two common calibration techniques are: Platt scaling (Sigmoid scaling): Suitable for correcting S-shaped calibration plots typically produced by models such as support vector machines (SVMs). Isotonic regression: Fits a non-decreasing step function to probabilities and is effective particularly with larger datasets, though it can sometimes lead to overfitting. Transforming predictions to frequentist probabilities is crucial as it provides probabilistic outputs that are directly interpretable as the actual likelihood of an event occurring. Such calibration significantly enhances the effectiveness of fixed position sizing methods, reducing maximum drawdowns and increasing risk-adjusted returns. However, calibration has less impact on position sizing methods that directly estimate parameters from the training data, such as ECDF and SOPS, suggesting that calibration is a critical step mainly for fixed methods that rely heavily on raw model outputs. =

Human–AI interaction

Human–AI interaction is a developing field of research and a sub-field of human–computer interaction (HCI). HCI is a field of research that explores the interactions between humans and computer-based technology, focusing on design implementation, user experience, and psychological factors. With the proliferation of artificial intelligence (AI), there has developed a sub-section of HCI research dedicated specifically to artificial intelligence and how people interact with and are impacted by it. This is human–AI interaction, abbreviated either as HAX or HAII. == Introduction == Artificial intelligence (AI), in general, has fluid definitions and varied research applications, but in brief can be applied to mechanizing tasks that would require human intelligence to complete. AI are tools designed to replicate the human abilities of navigating uncertainty, active learning, and processing information in different contexts. Within the context of HCI and HAX research, artificial intelligence can be broken into two sub-fields, natural language processing (NLP) and computer vision (CV). AI technologies notably include machine-learning, deep-learning and neural networks, and large-language models (LLMs). As a new and rapidly developing technology, AI is changing how computers work and therefore changing how humans interact with computers. Unlike the traditional human-computer interaction, where a human directs a machine, human-AI interaction is characterized by a more collaborative relationship between the computer program (the AI) and the human user, as AI is perceived as an active agent rather than a tool. This changing dynamic creates new questions and necessitates new research methods that are not present in traditional HCI research. According to a scoping review on the state of the discipline, the HAX field comprises research on the "design, development, and evaluation of AI systems" and encompasses the themes of human-AI collaboration, human-AI competition, human-AI conflict, and human-AI symbiosis. == Design == Machine learning and artificial intelligence have been used for decades in targeted advertising and to recommend content in social media. Ethical Guidelines (Framework for ethical AI development) == User Experience (UX) == This section should handle research on how users interact with tools. What techniques do they use, do they develop habits, what types of programs and devices are they using to access these tools, what do they use these tools to do exactly. === Cognitive Frameworks in AI Tool Users === AI has been viewed with various expectations, attributions, and often misconceptions. Many people exclusively understand AI as the LLM chatbots they interact with, like ChatGPT or Claude, or other generative AI programs. [Insert section: discuss how people interact with these specific AI tools as a connection to the following paragraphs] Most fundamentally, humans have a mental model of understanding AI's reasoning and motivation for its decision recommendations, and building a holistic and precise mental model of AI helps people create prompts to receive more valuable responses from AI. However, these mental models are not whole because people can only gain more information about AI through their limited interaction with it; more interaction with AI builds a better mental model that a person may build to produce better prompt outcomes. Research on human-AI interaction has emphasized that users develop mental models of AI systems and revise those models through repeated use, feedback, and explanation, while design research has stressed the importance of communicating capabilities and limitations early and supporting trust calibration through explanation and correction. In a 2025 SSRN working paper, John DeVadoss proposed "Hypothetico-Deductive Interaction" (HDI), a framework that describes human-AI interaction as a mutual process of conjecture and refutation in which users test assumptions about an AI system's capabilities while the system infers and updates assumptions about user goals through its responses and clarifying questions. DeVadoss argued that this framing helps explain prompt iteration, weak capability awareness, and trust miscalibration, and suggested design responses such as clearer communication of uncertainty, easier correction, actionable explanations, and safer failure modes. == Research themes == === Human-AI collaboration === Human-AI collaboration occurs when the human and AI supervise the task on the same level and extent to achieve the same goal. Some collaboration occurs in the form of augmenting human capability. AI may help human ability in analysis and decision-making through providing and weighing a volume of information, and learning to defer to the human decision when it recognizes its unreliability. It is especially beneficial when the human can detect a task that AI can be trusted to make few errors so that there is not a lot of excessive checking process required on the human's end. Some findings show signs of human-AI augmentation, or human–AI symbiosis, in which AI enhances human ability in a way that co-working on a task with AI produces better outcomes than a human working alone. For example: the quality and speed of customer service tasks increase when a human agent collaborates with AI, training on specific models allows AI to improve diagnoses in clinical settings, and AI with human-intervention can improve creativity of artwork while fully AI-generated haikus were rated negatively. Human-AI synergy, a concept in which human-AI collaboration would produce more optimal outcomes than either human or AI working alone could explain why AI does not always help with performance. Some AI features and development may accelerate human-AI synergy, while others may stagnate it. For example, when AI updates for better performance, it sometimes worsens the team performance with human and AI by reducing the compatibility with the new model and the mental model a user has developed on the previous version. Research has found that AI often supports human capabilities in the form of human-AI augmentation and not human-AI synergy, potentially because people rely too much on AI and stop thinking on their own. Prompting people to actively engage in analysis and think when to follow AI recommendations reduces their over-reliance, especially for individuals with higher need for cognition. === Human-AI competition === Robots and computers have substituted routine tasks historically completed by humans, but agentic AI has made it possible to also replace cognitive tasks including taking phone calls for appointments and driving a car. At the point of 2016, research has estimated that 45% of paid activities could be replaced by AI by 2030. Perceived autonomy of robots is known to increase people's negative attitude toward them, and worry about the technology taking over leads people to reject it. There has been a consistent tendency of algorithm aversion in which people prefer human advice over AI advice. However, people are not always able to tell apart tasks completed by AI or other humans. See AI takeover for more information. It is also notable that this sentiment is more prominent in the Western cultures as Westerners tend to show less positive views about AI compared to East Asians. == Research on the psychological impacts of AI == === Perception on others who use AI === As much as people perceive and make judgment about AI itself, they also form impressions of themselves and others who use AI. In the workplace, employees who disclose the use of AI in their tasks are more likely to receive feedback that they are not as hardworking as those who are in the same job who receive non-AI help to complete the same tasks. AI use disclosure diminishes the perceived legitimacy in the employee's task and decision making which ultimately leads observers to distrust people who use AI. Although these negative effects of AI use disclosure are weakened by the observers who use AI frequently themselves, the effect is still not attenuated by the observers' positive attitude towards AI. === Bias, AI, and human === Although AI provides a wide range of information and suggestions to its users, AI itself is not free of biases and stereotypes, and it does not always help people reduce their cognitive errors and biases. People are prone to such errors by failing to see other potential ideas and cases that are not listed by AI responses and committing to a decision suggested by AI that directly contradicts the correct information and directions that they are already aware of. Gender bias is also reflected as the female gendering of AI technologies which conceptualizes females as a helpful assistant. == Emotional connection with AI == Human-AI interaction has been theorized in the context of interpersonal relationships mainly in social psychology, communications and media studies, and as a technology interface through the lens of hu