{"id":9857,"date":"2020-09-25T08:07:55","date_gmt":"2020-09-25T08:07:55","guid":{"rendered":"https:\/\/www.experfy.com\/blog\/?p=9857"},"modified":"2023-10-30T13:32:57","modified_gmt":"2023-10-30T13:32:57","slug":"programming-fairness-in-algorithms","status":"publish","type":"post","link":"https:\/\/www.experfy.com\/blog\/ai-ml\/programming-fairness-in-algorithms\/","title":{"rendered":"Programming Fairness in Algorithms"},"content":{"rendered":"\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

Understanding and combating issues of fairness in supervised learning.<\/p>\n\n\n\n

\n

\u201cBeing good is easy, what is difficult is being just.\u201d\u00a0\u2015\u00a0<\/em>Victor Hugo<\/em><\/strong><\/p>\n

\u201cWe need to defend the interests of those whom we\u2019ve never met and never will.\u201d\u00a0\u2015\u00a0<\/em>Jeffrey D. Sachs<\/em><\/strong><\/p>\n<\/blockquote>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

Note:\u00a0<\/strong>This article is intended for a general audience to try and elucidate the complicated nature of unfairness in machine learning algorithms. As such, I have tried to explain concepts in an accessible way with minimal use of mathematics, in the hope that everyone can get something out of reading this.<\/p>\n\n\n\n

Supervised machine learning algorithms are inherently discriminatory. They are discriminatory in the sense that they use information embedded in the features of data to separate instances into distinct categories \u2014 indeed, this is their designated purpose in life. This is reflected in the name for these algorithms which are often referred to as discriminative algorithms (splitting data into categories), in contrast to generative algorithms (generating data from a given category). When we use supervised machine learning, this \u201cdiscrimination\u201d is used as an aid to help us categorize our data into distinct categories within the data distribution, as illustrated below.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"\"\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

Whilst this occurs when we apply discriminative algorithms \u2014 such as support vector machines, forms of parametric regression (e.g. vanilla linear regression), and non-parametric regression (e.g. random forest, neural networks, boosting) \u2014 to any dataset, the outcomes may not necessarily have any moral implications. For example, using last week\u2019s weather data to try and predict the weather tomorrow has no moral valence attached to it. However, when our dataset is based on information that describes people \u2014 individuals, either directly or indirectly, this can inadvertently result in discrimination on the basis of group affiliation.<\/p>\n\n\n\n

Clearly then, supervised learning is a dual-use technology. It can be used to our benefits, such as for information (e.g. predicting the weather) and protection (e.g. analyzing computer networks to detect attacks and malware). On the other hand, it has the potential to be weaponized to discriminate at essentially any level. This is not to say that the algorithms are evil for doing this, they are merely learning the representations present in the data, which may themselves have embedded within them the manifestations of historical injustices, as well as individual biases and proclivities. A common adage in data science is \u201cgarbage in = garbage out\u201d to refer to models being highly dependent on the quality of the data supplied to them. This can be stated analogously in the context of algorithmic fairness as \u201cbias in = bias out\u201d.<\/p>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Data Fundamentalism<\/strong><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

Some proponents believe in\u00a0data fundamentalism<\/strong><\/a>, that is to say, that the data reflects the objective truth of the world through empirical observations.<\/p>\n\n\n\n

\n

\u201cwith enough data, the numbers speak for themselves.\u201d\u00a0\u2014 Former Wired editor-in-chief Chris Anderson (a data fundamentalist)<\/em><\/strong><\/p>\n

Data and data sets are not objective; they are creations of human design. We give numbers their voice, draw inferences from them, and define their meaning through our interpretations. Hidden biases in both the collection and analysis stages present considerable risks, and are as important to the big-data equation as the numbers themselves.\u00a0\u2014 Kate Crawford, principal researcher at Microsoft Research Social Media Collective<\/em><\/strong><\/p>\n<\/blockquote>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n\n

Superficially, this seems like a reasonable hypothesis, but Kate Crawford provides a good counterargument in a\u00a0Harvard Business Review article<\/a>:<\/p>\n\n\n\n

\n

Boston has a problem with potholes, patching approximately 20,000 every year. To help allocate its resources efficiently, the City of Boston released the excellent\u00a0StreetBump smartphone app<\/a>, which draws on accelerometer and GPS data to help passively detect potholes, instantly reporting them to the city. While certainly a clever approach, StreetBump has a signal problem. People in lower income groups in the US are less likely to have smartphones, and this is particularly true of older residents, where smartphone penetration can be as low as 16%. For cities like Boston, this means that smartphone data sets are missing inputs from significant parts of the population \u2014 often those who have the fewest resources. \u2014\u00a0Kate Crawford, principal researcher at Microsoft Research<\/em><\/strong><\/p>\n<\/blockquote>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

Essentially, the StreetBump app picked up a preponderance of data from wealthy neighborhoods and relatively little from poorer neighborhoods. Naturally, the first conclusion you might draw from this is that the wealthier neighborhoods had more potholes, but in reality, there was just a lack of data from poorer neighborhoods because these people were less likely to have smartphones and thus have downloaded the SmartBump app. Often, it is data that we do not have in our dataset that can have the biggest impact on our results. This example illustrates a subtle form of discrimination on the basis of income. As a result, we should be cautious when drawing conclusions such as these from data that may suffer from a \u2018signal problem\u2019. This signal problem is often characterized as\u00a0sampling bias<\/strong><\/a>.<\/p>\n\n\n\n

Another notable example is the \u201cCorrectional Offender Management Profiling for Alternative Sanctions\u201d algorithm or COMPAS for short. This algorithm is used by a number of states across the United States to predict recidivism \u2014 the likelihood that a former criminal will re-offend. Analysis of this algorithm by ProPublica, an investigative journalism organization,\u00a0sparked controversy<\/a>\u00a0when it seemed to suggest that the algorithm was discriminating on the basis of race \u2014 a protected class in the United States. To give us a better idea of what is going on, the algorithm used to predict recidivism looks something like this:<\/p>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Recidivism Risk Score<\/strong>\u00a0= (age*\u2212w)+(age-at-first-arrest*\u2212w)+(history of violence*w) + (vocation education * w) + (history of noncompliance * w)<\/h3>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Do you have a feeling that deep learning on graphs is a bunch of heuristics that work sometimes and nobody has a clue why? In this post, I discuss the graph isomorphism problem, the Weisfeiler-Lehman heuristic for graph isomorphism testing, and how it can be used to analyse the expressive power of graph neural networks.<\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

It should be clear that race is not one of the variables used as a predictor. However, the data distribution between two given races may be significantly different for some of these variables, such as the \u2018history of violence\u2019 and \u2018vocation education\u2019 factors, based on historical injustices in the United States as well as demographic, social, and law enforcement statistics (which are often another target for criticism since they often use algorithms to determine which neighborhoods to patrol). The mismatch between these data distributions can be leveraged by an algorithm, leading to disparities between races and thus to some extent a result that is moderately biased towards or against certain races. These entrenched biases will then be operationalized by the algorithm and continue to persist as a result, leading to further injustices. This loop is essentially a\u00a0self-fulfilling prophecy<\/a>.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Historical Injustices \u2192 Training Data \u2192 Algorithmic Bias in Production<\/strong><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

This leads to some difficult questions \u2014 do we remove these problematic variables? How do we determine whether a feature will lead to discriminatory results? Do we need to engineer a metric that provides a threshold for \u2018discrimination\u2019? One could take this to the extreme and remove almost all variables, but then the algorithm would be of no use. This paints a bleak picture, but fortunately, there are ways to tackle these issues that will be discussed later in this article.<\/p>\n\n\n\n

These examples are not isolated incidents. Even breast cancer prediction algorithms show a level of unfair discrimination. Deep learning algorithms to predict breast cancer from mammograms are much less accurate for black women than white women. This is partly because the dataset used to train these algorithms is predominantly based on mammograms of white women, but also because the data distribution for breast cancer between black women and white women likely has substantial differences.\u00a0According to the Center for Disease Control (CDC)<\/a>\u00a0\u201cBlack women and white women get breast cancer at about the same rate, but black women die from breast cancer at a higher rate than white women<\/em>\u201d<\/strong>.<\/p>\n\n

\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

These issues raise questions about the motives of algorithmic developers \u2014 did the individuals that designed these models do so knowingly? Do they have an agenda they are trying to push and trying to hide it inside gray box machine learning models?<\/p>\n\n\n\n

Although these questions are impossible to answer with certainty, it is useful to consider Hanlon\u2019s razor when asking such questions:<\/p>\n\n\n\n

\n

Never attribute to malice that which is adequately explained by stupidity \u2014\u00a0Robert J. Hanlon<\/em><\/strong><\/p>\n<\/blockquote>\n\n\n\n

In other words, there are not that many evil people in the world (thankfully), and there are certainly less evil people in the world than there are incompetent people. On average, we should assume that when things go wrong it is more likely attributable to incompetence, naivety, or oversight than to outright malice. Whilst there are likely some malicious actors who would like to push discriminative agendas, these are likely a minority.<\/p>\n\n\n\n

Based on this assumption, what could have gone wrong? One could argue that statisticians, machine learning practitioners, data scientists, and computer scientists are not adequately taught how to develop supervised learning algorithms that control and correct for prejudicial proclivities.<\/p>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t

Why is this the case?<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

In truth, techniques that achieve this do not exist. Machine learning fairness is a young subfield of machine learning that has been growing in popularity over the last few years in response to the rapid integration of machine learning into social realms. Computer scientists, unlike doctors, are not necessarily trained to consider the ethical implications of their actions. It is only relatively recently (one could argue since the advent of social media) that the designs or inventions of computer scientists were able to take on an ethical dimension.<\/p>\n\n\n\n

This is demonstrated in the fact that most computer science journals do not require ethical statements or considerations for submitted manuscripts. If you take an image database full of millions of images of real people, this can without a doubt have ethical implications. By virtue of physical distance and the size of the dataset, computer scientists are so far removed from the data subjects that the implications on any one individual may be perceived as negligible and thus disregarded. In contrast, if a sociologist or psychologist performs a test on a small group of individuals, an entire ethical review board is set up to review and approve the experiment to ensure it does not transgress across any ethical boundaries.<\/p>\n\n\n\n

On the bright side, this is slowly beginning to change. More data science and computer science programs are starting to require students to take classes on data ethics and critical thinking, and journals are beginning to recognize that ethical reviews through IRBs and ethical statements in manuscripts may be a necessary addition to the peer-review process. The rising interest in the topic of machine learning<\/a> fairness is only strengthening this position.<\/p>\n\n\n

\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t

Fairness in Machine Learning<\/strong><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"\"\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

As mentioned previously, widespread adoption of supervised machine learning algorithms has raised concerns about algorithmic fairness. The more these algorithms are adopted, and the increasing control they have on our lives will only exacerbate these concerns. The machine learning community is well aware of these challenges and algorithmic fairness is now a rapidly developing subfield of machine learning with many excellent researchers such as Moritz Hardt, Cynthia Dwork, Solon Barocas, and Michael Feldman.<\/p>\n\n\n\n

That being said, there are still major hurdles to overcome before we can achieve truly fair algorithms. It is fairly easy to prevent\u00a0disparate treatment<\/strong>\u00a0in algorithms \u2014 the explicit differential treatment of one group over another, such as by removing variables that correspond to these attributes from the dataset (e.g. race, gender). However, it is much less easy to prevent\u00a0disparate impact<\/strong>\u00a0\u2014implicit differential treatment of one group over another, usually caused by something called\u00a0redundant encodings<\/em>\u00a0in the data.<\/p>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"\"\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\n

A\u00a0redundant encoding<\/strong>\u00a0tells us information about a protected attribute, such as race or gender, based on features present in our dataset that correlate with these attributes. For example, buying certain products online (such as makeup) may be highly correlated with gender, and certain zip codes may have different racial demographics that an algorithm might pick up on.<\/p>\n\n\n\n

Although an algorithm is not trying to discriminate along these lines, it is inevitable that data-driven algorithms that supersede human performance on pattern recognition tasks might pick up on these associations embedded within data, however small they may be. Additionally, if these associations were non-informative (i.e. they do not increase the accuracy of the algorithm) then the algorithm would ignore them, meaning that some information is clearly embedded in these protected attributes. This raises many challenges to researchers, such as:<\/p>\n\n\n\n