Software engineers make up a significant portion of the global workforce. As of 2022, there are an estimated 26.9 million professional software engineers worldwide, up from 21 million in 2016. == By country == === United States === In 2023, there were an estimated 1.6 million professional software developers in North America. There are 166 million people employed in the US workforce, making software developers 0.96% of the total workforce. ==== Summary ==== ==== Software engineers vs. traditional engineers ==== The following two tables compare the number of software engineers (611,900 in 2002) versus the number of traditional engineers (1,157,020 in 2002). There are another 1,500,000 people in system analysis, system administration, and computer support, many of whom might be called software engineers. Many systems analysts manage software development teams, and as analysis is an important software engineering role, many of them may be considered software engineers in the near future. This means that the number of software engineers may actually be much higher. It is important to note that the number of software engineers declined by 5 to 10 percent from 2000 to 2002. ==== Computer managers vs. construction and engineering managers ==== Computer and information system managers (264,790) manage software projects, as well as computer operations. Similarly, Construction and engineering managers (413,750) oversee engineering projects, manufacturing plants, and construction sites. Computer management is 64% the size of construction and engineering management. ==== Software engineering educators vs. engineering educators ==== Most people working in the field of computer science, whether making software systems (software engineering) or studying the theoretical and mathematical facts of software systems (computer science), acquire degrees in computer science. According to the U.S. Bureau of Labor Statistics (May 2023 data), there were approximately 44,800 postsecondary computer science teachers and 50,300 engineering teachers, indicating that the computer science educator workforce is nearly 89% as large as that of engineering educators. The combined number of postsecondary chemistry (25,400) and physics (17,100) teachers totaled 42,500, slightly less than the number of computer science educators. ==== Other software and engineering roles ==== ==== Relation to IT demographics ==== Software engineers are part of the much larger software, hardware, application, and operations community. In 2000 in the U.S., there were about 680,000 software engineers and about 10,000,000 IT workers. As of early 2025, there are an estimated 47.2 million software developers worldwide, representing a 50% increase from 31 million in Q1 2022. There are no numbers on testers in the BLS data. === India === There has been a healthy growth in the number of India's IT professionals over the past few years. From a base of 6,800 knowledge workers in 1985–86, the number increased to 522,000 software and services professionals by the end of 2001–02. It is estimated that out of these 528,000 knowledge workers, almost 170,000 are working in the IT software and services export industry; nearly 106,000 are working in the IT enabled services and over 230,000 in user organizations. === Australia === In May 2024, the Australian government reported that 169,300 Australians are employed as software and applications programmers, 17% of who are women. The role grew annually by 8,300 workers. === Russia === According to the Russian government, the number of IT specialists in the country increased by 13% in 2023, reaching approximately 857,000. During the initial phase of the 2022 invasion of Ukraine, an estimated 100,000 IT specialists left Russia.
Deep Learning Anti-Aliasing
Deep Learning Anti-Aliasing (DLAA) is a form of spatial anti-aliasing developed by Nvidia. DLAA depends on and requires Tensor Cores available in Nvidia RTX cards. DLAA is similar to Deep Learning Super Sampling (DLSS) in its anti-aliasing method, with one important differentiation being that the goal of DLSS is to increase performance at the cost of image quality, whereas the main priority of DLAA is improving image quality at the cost of performance (irrelevant of resolution upscaling or downscaling). DLAA is similar to temporal anti-aliasing (TAA) in that they are both spatial anti-aliasing solutions relying on past frame data. Compared to TAA, DLAA is substantially better when it comes to shimmering, flickering, and handling small meshes like wires. == Technical overview == DLAA collects game rendering data including raw low-resolution input, motion vectors, depth buffers, and exposure information. This information feeds into a convolutional neural network that processes the image to reduce aliasing while preserving fine detail. The neural network architecture employs an auto-encoder design trained on high-quality reference images. The training dataset includes diverse scenarios focusing on challenging cases like sub-pixel details, high-contrast edges, and transparent surfaces. The network then processes frames in real-time. Unlike traditional anti-aliasing solutions that rely on manually written heuristics, such as TAA, DLAA uses its neural network to preserve fine details while eliminating unwanted visual artifacts. == History == DLAA was initially called and marketed by Nvidia as DLSS 2x. The first game that added support for DLAA was The Elder Scrolls Online, which implemented the feature in 2021. By June 2022, DLAA was only available in six games. This number rose to 17 by February 2023. In June 2023, TechPowerUp reported that "DLAA is seeing sluggish adoption among game developers", and that Nvidia was working on adding DLAA to the quality presets of DLSS to boost adoption. By December 2023, DLAA was supported in 41 games. In early 2025, an update for the Nvidia App added a driver-based DLSS override feature that enables users to activate DLAA even in games that do not support it natively. == Differences between TAA and DLAA == TAA is used in many modern video games and game engines; however, all previous implementations have used some form of manually written heuristics to prevent temporal artifacts such as ghosting and flickering. One example of this is neighborhood clamping which forcefully prevents samples collected in previous frames from deviating too much compared to nearby pixels in newer frames. This helps to identify and fix many temporal artifacts, but deliberately removing fine details in this way is analogous to applying a blur filter, and thus the final image can appear blurry when using this method. DLAA uses an auto-encoder convolutional neural network trained to identify and fix temporal artifacts, instead of manually programmed heuristics as mentioned above. Because of this, DLAA can generally resolve detail better than other TAA and TAAU implementations, while also removing most temporal artifacts. == Differences between DLSS and DLAA == While DLSS handles upscaling with a focus on performance, DLAA handles anti-aliasing with a focus on visual quality. DLAA runs at the given screen resolution with no upscaling or downscaling functionality provided by DLAA. DLSS and DLAA share the same AI-driven anti-aliasing method. As such, DLAA functions like DLSS without the upscaling part. Both are made by Nvidia and require Tensor Cores. However, DLSS and DLAA cannot be enabled at the same time, only one can be selected depending on whether performance or image quality is prioritized. == Reception == TechPowerUp found that "[c]ompared to TAA and DLSS, DLAA is clearly producing the best image quality, especially at lower resolutions", arguing that, while "DLSS was already doing a better job than TAA at reconstructing small objects", "DLAA does an even better job". In a Cyberpunk 2077 performance test, IGN stated that "DLAA provided somewhat similar results [FPS wise] to the normal raster mode in most cases but got significant performance boost with the help of frame generation", a feature not available when using native resolution. Rock Paper Shotgun noted that, while DLAA is "not a completely perfect form of anti-aliasing, as the occasional jaggies are present", it "looks a lot sharper overall [than TAA], and especially in motion." According to PC World, "DLAA offers very good anti-aliasing without losing visual information — alternatives like TAA tend to struggle during motion-filled scenes, where DLAA doesn’t. Furthermore, DLAA’s loss of performance is lower than with conventional anti-aliasing methods."
Natarajan dimension
In the theory of Probably Approximately Correct Machine Learning, the Natarajan dimension characterizes the complexity of learning a set of functions, generalizing from the Vapnik–Chervonenkis dimension for boolean functions to multi-class functions. Originally introduced as the Generalized Dimension by Natarajan, it was subsequently renamed the Natarajan Dimension by Haussler and Long. == Definition == Let H {\displaystyle H} be a set of functions from a set X {\displaystyle X} to a set Y {\displaystyle Y} . H {\displaystyle H} shatters a set C ⊂ X {\displaystyle C\subset X} if there exist two functions f 0 , f 1 ∈ H {\displaystyle f_{0},f_{1}\in H} such that For every x ∈ C , f 0 ( x ) ≠ f 1 ( x ) {\displaystyle x\in C,f_{0}(x)\neq f_{1}(x)} . For every B ⊂ C {\displaystyle B\subset C} , there exists a function h ∈ H {\displaystyle h\in H} such that for all x ∈ B , h ( x ) = f 0 ( x ) {\displaystyle x\in B,h(x)=f_{0}(x)} and for all x ∈ C − B , h ( x ) = f 1 ( x ) {\displaystyle x\in C-B,h(x)=f_{1}(x)} . The Natarajan dimension of H is the maximal cardinality of a set shattered by H {\displaystyle H} . It is easy to see that if | Y | = 2 {\displaystyle |Y|=2} , the Natarajan dimension collapses to the Vapnik–Chervonenkis dimension. Shalev-Shwartz and Ben-David present comprehensive material on multi-class learning and the Natarajan dimension, including uniform convergence and learnability. Recently, Cohen et al showed that the Natarajan dimension is the dominant term governing agnostic multi-class PAC learnability.
Prescription monitoring program
In the United States, prescription monitoring programs (PMPs) or prescription drug monitoring programs (PDMPs) are state-run programs which collect and distribute data about the prescription and dispensation of federally controlled substances and, depending on state requirements, other potentially abusable prescription drugs. PMPs are meant to help prevent adverse drug-related events such as opioid overdoses, drug diversion, and substance abuse by decreasing the amount and/or frequency of opioid prescribing, and by identifying those patients who are obtaining prescriptions from multiple providers (i.e., "doctor shopping") or those physicians overprescribing opioids. Most US health care workers support the idea of PMPs, which intend to assist physicians, physician assistants, nurse practitioners, dentists and other prescribers, the pharmacists, chemists and support staff of dispensing establishments. The database, whose use is required by State law, typically requires prescribers and pharmacies dispensing controlled substances to register with their respective state PMPs and (for pharmacies and providers who dispense from their offices) to report the dispensation of such prescriptions to an electronic online database. The majority of PMPs are authorized to notify law enforcement agencies or licensing boards or physicians when a prescriber, or patients receiving prescriptions, exceed thresholds established by the state or prescription recipient exceeds thresholds established by the State. All states have implemented PDMPs, although evidence for the effectiveness of these programs is mixed. While prescription of opioids has decreased with PMP use, overdose deaths in many states have actually increased, with those states sharing data with neighboring jurisdictions or requiring reporting of more drugs experiencing highest increases in deaths. This may be because those declined opioid prescriptions turn to street drugs, whose potency and contaminants carry greater overdose risk. == History == Prescription drug monitoring programs, or PDMPs, are an example of one initiative proposed to alleviate effects of the opioid crisis. The programs are designed to restrict prescription drug abuse by limiting a patient's ability to obtain similar prescriptions from multiple providers (i.e. “doctor shopping”) and reducing diversion of controlled substances. This is meant to reduce risk of fatal overdose caused by high doses of opioids or interactions between opioids and benzodiazepenes, and to enable better decision making on the part of healthcare providers who may be unaware of a patient's prescription drug use, history or other prescriptions. PDMPs have been implemented in state legislations since 1939 in California, a time before electronic medical records, though implementation rose alongside increased awareness of overprescribing of opioids and overdose. A later New York state program was struck down by the U.S. Supreme Court in Whalen v. Roe. But, by 2019, 49 states, the District of Columbia, and Guam had enacted PDMP legislation. In 2021 Missouri, the last State to not use a PMP, adopted legislation to create one. PMPs are constantly being updated to increase speed of data collection, sharing of data across States, and ease of interpretation. This is being done by integrating PDMP reports with other health information technologies such as health information exchanges (HIE), electronic health record (EHR) systems, and/ or pharmacy dispensing software systems. One program that has been implemented in nine states is called the PDMP Electronic Health Records Integration and Interoperability Expansion, also known as PEHRIIE. Another software, marketed by Bamboo Health and integrated with PMPs in 43 states, uses an algorithm to track factors thought to increase risk of diversion, abuse or overdose, and assigns patients a three digit score based on presumed indicators of risk. While some studies have suggested that PDMP-HIT integration and sharing of interstate data brings benefits such as reduced opioid-related inpatient morbidity, others have found no or negative impact on mortality compared to states without PMP data sharing. Patient and media reports suggest need for testing and evaluation of algorithmic software used to score risk, with some patients reporting denial of prescriptions without c explanation or clarity of data. == Goals == Most health care workers support PMPs which intend to assist physicians, physician assistants, nurse practitioners, dentists and other prescribers, the pharmacists, chemists and support staff of dispensing establishments, as well as law-enforcement agencies. The collaboration supports the legitimate medical use of controlled substances while limiting their abuse and diversion. Pharmacies dispensing controlled substances and prescribers typically must register with their respective state PMPs and (for pharmacies and providers who dispense controlled substances from their offices) report the dispensation to an electronic online database. Some pharmacy software can submit these reports automatically to multiple states. == Usage == === List of programs by state === === Software systems === NarxCare is a prescription drug monitoring program (PDMP) run by Bamboo Health. Bamboo Health was formerly known as Appriss. It is widely used across the United States by pharmacies including Rite Aid as well as those at Walmart and Sam’s Club. The NarxCare software allows doctors to view data about a patient, combining data from the prescription registries of various U.S. states to make the registries interoperable nationally. It also uses machine learning to generate an "Overdose Risk Score" that potentially includes EMS and criminal justice data; these scores have been criticized by researchers and patient advocates for the lack of transparency in the process as well as the potential for disparate treatment of women and minority groups. Advertised as an "analytics tool and care management platform", the NarxCare software allows doctors to view data about a patient including how many pharmacies they have visited and the combinations of medication they are prescribed. It combines data from the prescription registries of various U.S. states, making the registries interoperable nationally. It additionally uses machine learning to generate various three-digit "risk scores" and an overall "Overdose Risk Score", collectively referred to as Narx Scores, in a process that potentially includes EMS and criminal justice data as well as court records. == Controversy == Many doctors and researchers support the idea of PDMPs as a tool in combatting the opioid epidemic. Opioid prescribing, opioid diversion and supply, opioid misuse, and opioid-related morbidity and mortality are common elements in data entered into PDMPs. Prescription Monitoring Programs are purported to offer economic benefits for the states who implement them by decreasing overall health care costs, lost productivity, and investigation times. However, there are many studies that conclude the impact of PDMPs is unclear. While use of PMPs has been accompanied by decrease in opioid prescribing, few analyses consider corresponding use of street opioids, extramedical use, or diversion, which might provide a more holistic method for evaluation of PMP intent and efficacy. Evidence for PDMP impact on fatal overdoses is decidedly mixed, with multiple studies finding increased overdose rates in some states, decreases in others, or no clear impact. Interestingly, an increase in heroin overdoses after PDMP implementation has been commonly reported, presumably as denial of prescription opioids sends patients in search of street drugs. Narx Scores have been criticized by researchers and patient advocates for the lack of transparency in the generation process as well as the potential for disparate treatment of women and minority groups. Writing in Duke Law Journal, Jennifer Oliva stated that "black-box algorithms" are used to generate the scores.
CN2 algorithm
The CN2 induction algorithm is a learning algorithm for rule induction. It is designed to work even when the training data is imperfect. It is based on ideas from the AQ algorithm and the ID3 algorithm. As a consequence it creates a rule set like that created by AQ but is able to handle noisy data like ID3. == Description of algorithm == The algorithm must be given a set of examples, TrainingSet, which have already been classified in order to generate a list of classification rules. A set of conditions, SimpleConditionSet, which can be applied, alone or in combination, to any set of examples is predefined to be used for the classification. routine CN2(TrainingSet) let the ClassificationRuleList be empty repeat let the BestConditionExpression be Find_BestConditionExpression(TrainingSet) if the BestConditionExpression is not nil then let the TrainingSubset be the examples covered by the BestConditionExpression remove from the TrainingSet the examples in the TrainingSubset let the MostCommonClass be the most common class of examples in the TrainingSubset append to the ClassificationRuleList the rule 'if ' the BestConditionExpression ' then the class is ' the MostCommonClass until the TrainingSet is empty or the BestConditionExpression is nil return the ClassificationRuleList routine Find_BestConditionExpression(TrainingSet) let the ConditionalExpressionSet be empty let the BestConditionExpression be nil repeat let the TrialConditionalExpressionSet be the set of conditional expressions, {x and y where x belongs to the ConditionalExpressionSet and y belongs to the SimpleConditionSet}. remove all formulae in the TrialConditionalExpressionSet that are either in the ConditionalExpressionSet (i.e., the unspecialized ones) or null (e.g., big = y and big = n) for every expression, F, in the TrialConditionalExpressionSet if F is statistically significant and F is better than the BestConditionExpression by user-defined criteria when tested on the TrainingSet then replace the current value of the BestConditionExpression by F while the number of expressions in the TrialConditionalExpressionSet > user-defined maximum remove the worst expression from the TrialConditionalExpressionSet let the ConditionalExpressionSet be the TrialConditionalExpressionSet until the ConditionalExpressionSet is empty return the BestConditionExpression
Color normalization
Color normalization is a topic in computer vision concerned with artificial color vision and object recognition. In general, the distribution of color values in an image depends on the illumination, which may vary depending on lighting conditions, cameras, and other factors. Color normalization allows for object recognition techniques based on color to compensate for these variations. == Main concepts == === Color constancy === Color constancy is a feature of the human internal model of perception, which provides humans with the ability to assign a relatively constant color to objects even under different illumination conditions. This is helpful for object recognition as well as identification of light sources in an environment. For example, humans see an object approximately as the same color when the sun is bright or when the sun is dim. === Applications === Color normalization has been used for object recognition on color images in the field of robotics, bioinformatics and general artificial intelligence, when it is important to remove all intensity values from the image while preserving color values. One example is in case of a scene shot by a surveillance camera over the day, where it is important to remove shadows or lighting changes on same color pixels and recognize the people that passed. Another example is automated screening tools used for the detection of diabetic retinopathy as well as molecular diagnosis of cancer states, where it is important to include color information during classification. == Known issues == The main issue about certain applications of color normalization is that the result looks unnatural or too distant from the original colors. In cases where there is a subtle variation between important aspects, this can be problematic. More specifically, the side effect can be that pixels become divergent and not reflect the actual color value of the image. A way of combating this issue is to use color normalization in combination with thresholding to correctly and consistently segment a colored image. == Transformations and algorithms == There is a vast array of different transformations and algorithms for achieving color normalization and a limited list is presented here. The performance of an algorithm is dependent on the task and one algorithm which performs better than another in one task might perform worse in another (no free lunch theorem). Additionally, the choice of the algorithm depends on the preferences of the user for the end-result, e.g. they may want a more natural-looking color image. === Grey world === The grey world normalization makes the assumption that changes in the lighting spectrum can be modelled by three constant factors applied to the red, green and blue channels of color. More specifically, a change in illuminated color can be modelled as a scaling α, β and γ in the R, G and B color channels and as such the grey world algorithm is invariant to illumination color variations. Therefore, a constancy solution can be achieved by dividing each color channel by its average value as shown in the following formula: ( α R , β G , γ B ) → ( α R α n ∑ i R , β G β n ∑ i G , γ B γ n ∑ i B ) {\displaystyle \left(\alpha R,\beta G,\gamma B\right)\rightarrow \left({\frac {\alpha R}{{\frac {\alpha }{n}}\sum _{i}R}},{\frac {\beta G}{{\frac {\beta }{n}}\sum _{i}G}},{\frac {\gamma B}{{\frac {\gamma }{n}}\sum _{i}B}}\right)} As mentioned above, grey world color normalization is invariant to illuminated color variations α, β and γ, however it has one important problem: it does not account for all variations of illumination intensity and it is not dynamic; when new objects appear in the scene it fails. To solve this problem there are several variants of the grey world algorithm. Additionally there is an iterative variation of the grey world normalization, however it was not found to perform significantly better. === Histogram equalization === Histogram equalization is a non-linear transform which maintains pixel rank and is capable of normalizing for any monotonically increasing color transform function. It is considered to be a more powerful normalization transformation than the grey world method. The results of histogram equalization tend to have an exaggerated blue channel and look unnatural, due to the fact that in most images the distribution of the pixel values is usually more similar to a Gaussian distribution, rather than uniform. === Histogram specification === Histogram specification transforms the red, green and blue histograms to match the shapes of three specific histograms, rather than simply equalizing them. It refers to a class of image transforms which aims to obtain images of which the histograms have a desired shape. As specified, firstly it is necessary to convert the image so that it has a particular histogram. Assume an image x. The following formula is the equalization transform of this image: y = f ( x ) = ∫ 0 x p x ( u ) d u {\displaystyle y=f(x)=\int \limits _{0}^{x}p_{x}(u)du} Then assume wanted image z. The equalization transform of this image is: y ′ = g ( z ) = ∫ 0 z p z ( u ) d u {\displaystyle y'=g(z)=\int \limits _{0}^{z}p_{z}(u)du} Of course p z ( u ) {\displaystyle p_{z}(u)} is the histogram of the output image. The formula to find the inverse of the above transform is: z = g − 1 ( y ′ ) {\displaystyle z=g^{-1}(y')} Therefore, since images y and y' have the same equalized histogram they are actually the same image, meaning y = y' and the transform from the given image x to the wanted image z is: z = g − 1 ( y ′ ) = g − 1 ( y ) = g − 1 ( f ( x ) ) {\displaystyle z=g^{-1}(y')=g^{-1}(y)=g^{-1}(f(x))} Histogram specification has the advantage of producing more realistic looking images, as it does not exaggerate the blue channel like histogram equalization. === Comprehensive Color Normalization === The comprehensive color normalization is shown to increase localization and object classification results in combination with color indexing. It is an iterative algorithm which works in two stages. The first stage is to use the red, green and blue color space with the intensity normalized, to normalize each pixel. The second stage is to normalize each color channel separately, so that the sum of the color components is equal to one third of the number of pixels. The iterations continue until convergence, meaning no additional changes. Formally: Normalize the color image f ( t ) = [ f i j ( t ) ] i = 1... N , j = 1... M {\displaystyle f^{(t)}=[f_{ij}^{(t)}]_{i=1...N,j=1...M}} which consists of color vectors f i j ( t ) = ( r i j ( t ) , g i j ( t ) , b i j ( t ) ) T . {\displaystyle f_{ij}^{(t)}=(r_{ij}^{(t)},g_{ij}^{(t)},b_{ij}^{(t)})^{T}.} For the first step explained above, compute: S i j := r i j ( t ) + g i j ( t ) + b i j ( t ) {\displaystyle S_{ij}:=r_{ij}^{(t)}+g_{ij}^{(t)}+b_{ij}^{(t)}} which leads to r i j ( t + 1 ) = r i j ( t ) S i j , g i j ( t + 1 ) = g i j ( t ) S i j {\displaystyle r_{ij}^{(t+1)}={\frac {r_{ij}^{(t)}}{S_{ij}}},g_{ij}^{(t+1)}={\frac {g_{ij}^{(t)}}{S_{ij}}}} and b i j ( t + 1 ) = b i j ( t ) S i j . {\displaystyle b_{ij}^{(t+1)}={\frac {b_{ij}^{(t)}}{S_{ij}}}.} For the second step explained above, compute: r ′ = 3 N M ∑ i = 1 N ∑ j = 1 M r i j ( t + 1 ) {\displaystyle r'={\frac {3}{NM}}\sum _{i=1}^{N}\sum _{j=1}^{M}r_{ij}^{(t+1)}} and normalize r i j ( t + 2 ) = r i j ( t + 1 ) r ′ . {\displaystyle r_{ij}^{(t+2)}={\frac {r_{ij}^{(t+1)}}{r'}}.} Of course the same process is done for b' and g'. Then these two steps are repeated until the changes between iteration t and t+2 are less than some set threshold. Comprehensive color normalization, just like the histogram equalization method previously mentioned, produces results that may look less natural due to the reduction in the number of color values.
Feature selection
In machine learning, feature selection is the process of selecting a subset of relevant features (variables, predictors) for use in model construction. Feature selection techniques are used for several reasons: simplification of models to make them easier to interpret, shorter training times, to avoid the curse of dimensionality, improve the compatibility of the data with a certain learning model class, to encode inherent symmetries present in the input space. The central premise when using feature selection is that data sometimes contains features that are redundant or irrelevant, and can thus be removed without incurring much loss of information. Redundancy and irrelevance are two distinct notions, since one relevant feature may be redundant in the presence of another relevant feature with which it is strongly correlated. Feature extraction creates new features from functions of the original features, whereas feature selection finds a subset of the features. Feature selection techniques are often used in domains where there are many features and comparatively few samples (data points). == Introduction == A feature selection algorithm can be seen as the combination of a search technique for proposing new feature subsets, along with an evaluation measure which scores the different feature subsets. The simplest algorithm is to test each possible subset of features finding the one which minimizes the error rate. This is an exhaustive search of the space, and is computationally intractable for all but the smallest of feature sets. The choice of evaluation metric heavily influences the algorithm, and it is these evaluation metrics which distinguish between the three main categories of feature selection algorithms: wrappers, filters and embedded methods. Wrapper methods use a predictive model to score feature subsets. Each new subset is used to train a model, which is tested on a hold-out set. Counting the number of mistakes made on that hold-out set (the error rate of the model) gives the score for that subset. As wrapper methods train a new model for each subset, they are very computationally intensive, but usually provide the best performing feature set for that particular type of model or typical problem. Filter methods use a proxy measure instead of the error rate to score a feature subset. This measure is chosen to be fast to compute, while still capturing the usefulness of the feature set. Common measures include the mutual information, the pointwise mutual information, Pearson product-moment correlation coefficient, Relief-based algorithms, and inter/intra class distance or the scores of significance tests for each class/feature combinations. Filters are usually less computationally intensive than wrappers, but they produce a feature set which is not tuned to a specific type of predictive model. This lack of tuning means a feature set from a filter is more general than the set from a wrapper, usually giving lower prediction performance than a wrapper. However the feature set doesn't contain the assumptions of a prediction model, and so is more useful for exposing the relationships between the features. Many filters provide a feature ranking rather than an explicit best feature subset, and the cut off point in the ranking is chosen via cross-validation. Filter methods have also been used as a preprocessing step for wrapper methods, allowing a wrapper to be used on larger problems. One other popular approach is the Recursive Feature Elimination algorithm, commonly used with Support Vector Machines to repeatedly construct a model and remove features with low weights. Embedded methods are a catch-all group of techniques which perform feature selection as part of the model construction process. The exemplar of this approach is the LASSO method for constructing a linear model, which penalizes the regression coefficients with an L1 penalty, shrinking many of them to zero. Any features which have non-zero regression coefficients are 'selected' by the LASSO algorithm. Improvements to the LASSO include Bolasso which bootstraps samples; Elastic net regularization, which combines the L1 penalty of LASSO with the L2 penalty of ridge regression; and FeaLect which scores all the features based on combinatorial analysis of regression coefficients. AEFS further extends LASSO to nonlinear scenario with autoencoders. These approaches tend to be between filters and wrappers in terms of computational complexity. In traditional regression analysis, the most popular form of feature selection is stepwise regression, which is a wrapper technique. It is a greedy algorithm that adds the best feature (or deletes the worst feature) at each round. The main control issue is deciding when to stop the algorithm. In machine learning, this is typically done by cross-validation. In statistics, some criteria are optimized. This leads to the inherent problem of nesting. More robust methods have been explored, such as branch and bound and piecewise linear network. == Subset selection == Subset selection evaluates a subset of features as a group for suitability. Subset selection algorithms can be broken up into wrappers, filters, and embedded methods. Wrappers use a search algorithm to search through the space of possible features and evaluate each subset by running a model on the subset. Wrappers can be computationally expensive and have a risk of over fitting to the model. Filters are similar to wrappers in the search approach, but instead of evaluating against a model, a simpler filter is evaluated. Embedded techniques are embedded in, and specific to, a model. Many popular search approaches use greedy hill climbing, which iteratively evaluates a candidate subset of features, then modifies the subset and evaluates if the new subset is an improvement over the old. Evaluation of the subsets requires a scoring metric that grades a subset of features. Exhaustive search is generally impractical, so at some implementor (or operator) defined stopping point, the subset of features with the highest score discovered up to that point is selected as the satisfactory feature subset. The stopping criterion varies by algorithm; possible criteria include: a subset score exceeds a threshold, a program's maximum allowed run time has been surpassed, etc. Alternative search-based techniques are based on targeted projection pursuit which finds low-dimensional projections of the data that score highly: the features that have the largest projections in the lower-dimensional space are then selected. Search approaches include: Exhaustive Best first Simulated annealing Genetic algorithm Greedy forward selection Greedy backward elimination Particle swarm optimization Targeted projection pursuit Scatter search Variable neighborhood search Two popular filter metrics for classification problems are correlation and mutual information, although neither are true metrics or 'distance measures' in the mathematical sense, since they fail to obey the triangle inequality and thus do not compute any actual 'distance' – they should rather be regarded as 'scores'. These scores are computed between a candidate feature (or set of features) and the desired output category. There are, however, true metrics that are a simple function of the mutual information; see here. Other available filter metrics include: Class separability Error probability Inter-class distance Probabilistic distance Entropy Consistency-based feature selection Correlation-based feature selection == Optimality criteria == The choice of optimality criteria is difficult as there are multiple objectives in a feature selection task. Many common criteria incorporate a measure of accuracy, penalised by the number of features selected. Examples include Akaike information criterion (AIC) and Mallows's Cp, which have a penalty of 2 for each added feature. AIC is based on information theory, and is effectively derived via the maximum entropy principle. Other criteria are Bayesian information criterion (BIC), which uses a penalty of log n {\displaystyle {\sqrt {\log {n}}}} for each added feature, minimum description length (MDL) which asymptotically uses log n {\displaystyle {\sqrt {\log {n}}}} , Bonferroni / RIC which use 2 log p {\displaystyle {\sqrt {2\log {p}}}} , maximum dependency feature selection, and a variety of new criteria that are motivated by false discovery rate (FDR), which use something close to 2 log p q {\displaystyle {\sqrt {2\log {\frac {p}{q}}}}} . A maximum entropy rate criterion may also be used to select the most relevant subset of features. == Structure learning == Filter feature selection is a specific case of a more general paradigm called structure learning. Feature selection finds the relevant feature set for a specific target variable whereas structure learning finds the relationships between all the variables, usually by expressing these relationships as a graph. The most common structure learning algorithms