{"id":8956,"date":"2021-03-16T20:10:18","date_gmt":"2021-03-17T03:10:18","guid":{"rendered":"https:\/\/www.hmc.edu\/about-hmc\/?p=8956"},"modified":"2021-03-22T16:31:33","modified_gmt":"2021-03-22T23:31:33","slug":"cs-research-published","status":"publish","type":"post","link":"https:\/\/www.hmc.edu\/about\/2021\/03\/16\/cs-research-published\/","title":{"rendered":"HMC CS Researchers Publish Chapter on Algorithmic Biases"},"content":{"rendered":"

Harvey Mudd College computer science professor George Monta\u00f1ez and his students Daniel Bashir \u201920 and Julius Lauw \u201920 have published the chapter \u201cTrading Bias for Expressivity in Artificial Learning\u201d in\u00a0ICAART 2020: Agents and Artificial\u00a0Intelligence<\/em>, part of the Lecture Notes in Computer Science book series (Springer, Cham).<\/p>\n

The HMC researchers\u2019 chapter about how bias relates to algorithm flexibility (expressivity) was an expanded and completely rewritten version of the lab\u2019s award-winning 2020 paper for the International Conference on Agents and Artificial Intelligence (ICAART).<\/p>\n

Monta\u00f1ez, Bashir and Lauw expanded their original paper by beginning with a definition of the term \u201cbias.\u201d<\/p>\n

\u201cThe word \u2018bias\u2019 is a loaded term in machine learning and statistics, with at least four different uses,\u201d says Monta\u00f1ez. \u201cWe added a section differentiating the meanings of the term and showing how our particular notion of bias, \u2018algorithmic bias,\u2019 is not equivalent to the prejudicial biases we rightly try to eliminate in data science. While all prejudicial biases create algorithmic bias, not all algorithmic biases are prejudicial.\u201d<\/p>\n

The authors also took advantage of having more time with their research to improve their presentation of the paper\u2019s core ideas. \u201cOften when you present a paper, in having to communicate the ideas simply to an audience, you stumble upon a much better way of presenting your work,\u201d Monta\u00f1ez says.<\/p>\n

\u201cAlthough all of the theorems and definitions are equivalent between the original paper and book chapter,\u201d he explains, \u201cthe extended version in the book introduces all of the key concepts around a geometric idea called\u00a0inductive orientation, which is basically a direction an algorithm \u2018points towards\u2019 in high-dimensional space. The degree to which it points somewhere away from the baseline direction is the degree to which it can be algorithmically biased\u2014we\u2019re basically measuring how well-aligned an algorithm is with regard to a particular situation we care about. Furthermore, pointing towards one direction means pointing away from other directions, so we see that no algorithm can be well-aligned with all situations. This geometric idea of alignment paints a better intuitive picture of what we mean by algorithmic biases.\u201d<\/p>\n

The original paper, \u201cThe Bias-Expressivity Trade-off,\u201d co-authored by Monta\u00f1ez, Lauw, Dominique Macias \u201919, Akshay Trikha \u201921 and Julia Vendemiatti \u201921, won the Best Paper award at ICAART 2020.<\/p>\n

\u201cThis chapter stands essentially as a new paper, which builds on the content of the original conference publication, but improves it in many ways,\u201d says Monta\u00f1ez. \u201cWe were also fortunate to have Daniel Bashir join us as a co-author; he was responsible for many of the improvements in the new work, including the new section on different biases in artificial learning.\u201d<\/p>\n

This publication marks Bashir\u2019s\u00a0third and Lauw\u2019s fifth with Monta\u00f1ez\u2019s AMISTAD Lab. In 2020, Lauw received a student researcher award from the Computer Science Department. Bashir was a 2020 CRA Outstanding Undergraduate Researcher honorable mention.<\/p>\n

\u201cThe chapter will likely be used by machine learning and AI practitioners who are interested in new ways of looking at and measuring biases in artificial learning systems,\u201d says Monta\u00f1ez. \u201cHopefully it inspires greater transparency concerning the biases present in all learning algorithms.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"

Harvey Mudd College computer science professor George Monta\u00f1ez and his students Daniel Bashir \u201920 and Julius Lauw \u201920 have published […]<\/p>\n","protected":false},"author":145,"featured_media":8958,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[11,14,26,30],"class_list":["post-8956","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-computer-science","category-faculty","category-research","category-students"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/posts\/8956","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/users\/145"}],"replies":[{"embeddable":true,"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/comments?post=8956"}],"version-history":[{"count":0,"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/posts\/8956\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/media\/8958"}],"wp:attachment":[{"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/media?parent=8956"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.hmc.edu\/about\/wp-json\/wp\/v2\/categories?post=8956"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}