Debjani Saha, Candice Schumann, Duncan C. McElfresh, John P. Dickerson, Michelle L. Mazurek, Michael Carl Tschantz
Bias in machine learning has manifested injustice in several areas, with notable examples including gender bias in job-related ads [4], racial bias in evaluating names on resumes [3], and racial bias in predicting criminal recidivism [1]. In response, research into algorithmic fairness has grown in both importance and volume over the past few years. Different metrics and approaches to algorithmic fairness have been proposed, many of which are based on prior legal and philosophical concepts [2]. The rapid expansion of this field makes it difficult for professionals to keep up, let alone the general public. Furthermore, misinformation about notions of fairness can have significant legal implications. Computer scientists have largely focused on developing mathematical notions of fairness and incorporating them in fielded ML systems. A much smaller collection of studies has measured public perception of bias and (un)fairness in algorithmic decision-making. However, one major question underlying the study of ML fairness remains unanswered in the literature: Does the general public understand mathematical definitions of ML fairness and their behavior in ML applications? We take a first step towards answering this question by studying non-expert comprehension and perceptions of one popular definition of ML fairness, demographic parity [5]. Specifically, we developed an online survey to address the following: (1) Does a non-technical audience comprehend the definition and implications of demographic parity? (2) Do demographics play a role in comprehension? (3) How are comprehension and sentiment related? (4) Does the application scenario affect comprehension?
{"title":"Human Comprehension of Fairness in Machine Learning","authors":"Debjani Saha, Candice Schumann, Duncan C. McElfresh, John P. Dickerson, Michelle L. Mazurek, Michael Carl Tschantz","doi":"10.1145/3375627.3375819","DOIUrl":"https://doi.org/10.1145/3375627.3375819","url":null,"abstract":"Bias in machine learning has manifested injustice in several areas, with notable examples including gender bias in job-related ads [4], racial bias in evaluating names on resumes [3], and racial bias in predicting criminal recidivism [1]. In response, research into algorithmic fairness has grown in both importance and volume over the past few years. Different metrics and approaches to algorithmic fairness have been proposed, many of which are based on prior legal and philosophical concepts [2]. The rapid expansion of this field makes it difficult for professionals to keep up, let alone the general public. Furthermore, misinformation about notions of fairness can have significant legal implications. Computer scientists have largely focused on developing mathematical notions of fairness and incorporating them in fielded ML systems. A much smaller collection of studies has measured public perception of bias and (un)fairness in algorithmic decision-making. However, one major question underlying the study of ML fairness remains unanswered in the literature: Does the general public understand mathematical definitions of ML fairness and their behavior in ML applications? We take a first step towards answering this question by studying non-expert comprehension and perceptions of one popular definition of ML fairness, demographic parity [5]. Specifically, we developed an online survey to address the following: (1) Does a non-technical audience comprehend the definition and implications of demographic parity? (2) Do demographics play a role in comprehension? (3) How are comprehension and sentiment related? (4) Does the application scenario affect comprehension?","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"69 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78540446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
AJ Alvero, Noah Arthurs, A. Antonio, B. Domingue, Ben Gebre-Medhin, Sonia Giebel, M. Stevens
College admissions in the United States is carried out by a human-centered method of evaluation known as holistic review, which typically involves reading original narrative essays submitted by each applicant. The legitimacy and fairness of holistic review, which gives human readers significant discretion over determining each applicant's fitness for admission, has been repeatedly challenged in courtrooms and the public sphere. Using a unique corpus of 283,676 application essays submitted to a large, selective, state university system between 2015 and 2016, we assess the extent to which applicant demographic characteristics can be inferred from application essays. We find a relatively interpretable classifier (logistic regression) was able to predict gender and household income with high levels of accuracy. Findings suggest that data auditing might be useful in informing holistic review, and perhaps other evaluative systems, by checking potential bias in human or computational readings.
{"title":"AI and Holistic Review: Informing Human Reading in College Admissions","authors":"AJ Alvero, Noah Arthurs, A. Antonio, B. Domingue, Ben Gebre-Medhin, Sonia Giebel, M. Stevens","doi":"10.1145/3375627.3375871","DOIUrl":"https://doi.org/10.1145/3375627.3375871","url":null,"abstract":"College admissions in the United States is carried out by a human-centered method of evaluation known as holistic review, which typically involves reading original narrative essays submitted by each applicant. The legitimacy and fairness of holistic review, which gives human readers significant discretion over determining each applicant's fitness for admission, has been repeatedly challenged in courtrooms and the public sphere. Using a unique corpus of 283,676 application essays submitted to a large, selective, state university system between 2015 and 2016, we assess the extent to which applicant demographic characteristics can be inferred from application essays. We find a relatively interpretable classifier (logistic regression) was able to predict gender and household income with high levels of accuracy. Findings suggest that data auditing might be useful in informing holistic review, and perhaps other evaluative systems, by checking potential bias in human or computational readings.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78869068","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Sandhya Saisubramanian, Sainyam Galhotra, S. Zilberstein
Graph clustering groups entities -- the vertices of a graph -- based on their similarity, typically using a complex distance function over a large number of features. Successful integration of clustering approaches in automated decision-support systems hinges on the interpretability of the resulting clusters. This paper addresses the problem of generating interpretable clusters, given features of interest that signify interpretability to an end-user, by optimizing interpretability in addition to common clustering objectives. We propose a β-interpretable clustering algorithm that ensures that at least β fraction of nodes in each cluster share the same feature value. The tunable parameter β is user-specified. We also present a more efficient algorithm for scenarios with β!=!1$ and analyze the theoretical guarantees of the two algorithms. Finally, we empirically demonstrate the benefits of our approaches in generating interpretable clusters using four real-world datasets. The interpretability of the clusters is complemented by generating simple explanations denoting the feature values of the nodes in the clusters, using frequent pattern mining.
{"title":"Balancing the Tradeoff Between Clustering Value and Interpretability","authors":"Sandhya Saisubramanian, Sainyam Galhotra, S. Zilberstein","doi":"10.1145/3375627.3375843","DOIUrl":"https://doi.org/10.1145/3375627.3375843","url":null,"abstract":"Graph clustering groups entities -- the vertices of a graph -- based on their similarity, typically using a complex distance function over a large number of features. Successful integration of clustering approaches in automated decision-support systems hinges on the interpretability of the resulting clusters. This paper addresses the problem of generating interpretable clusters, given features of interest that signify interpretability to an end-user, by optimizing interpretability in addition to common clustering objectives. We propose a β-interpretable clustering algorithm that ensures that at least β fraction of nodes in each cluster share the same feature value. The tunable parameter β is user-specified. We also present a more efficient algorithm for scenarios with β!=!1$ and analyze the theoretical guarantees of the two algorithms. Finally, we empirically demonstrate the benefits of our approaches in generating interpretable clusters using four real-world datasets. The interpretability of the clusters is complemented by generating simple explanations denoting the feature values of the nodes in the clusters, using frequent pattern mining.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"54 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81039764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Spencer Frazier, Md Sultan Al Nahian, Mark O. Riedl, Brent Harrison
Value alignment is a property of an intelligent agent indicating that it can only pursue goals and activities that are beneficial to humans. Traditional approaches to value alignment use imitation learning or preference learning to infer the values of humans by observing their behavior. We introduce a complementary technique in which a value-aligned prior is learned from naturally occurring stories which encode societal norms. Training data is sourced from the children's educational comic strip, Goofus & Gallant. In this work, we train multiple machine learning models to classify natural language descriptions of situations found in the comic strip as normative or non-normative by identifying if they align with the main characters' behavior. We also report the models' performance when transferring to two unrelated tasks with little to no additional training on the new task.
{"title":"Learning Norms from Stories: A Prior for Value Aligned Agents","authors":"Spencer Frazier, Md Sultan Al Nahian, Mark O. Riedl, Brent Harrison","doi":"10.1145/3375627.3375825","DOIUrl":"https://doi.org/10.1145/3375627.3375825","url":null,"abstract":"Value alignment is a property of an intelligent agent indicating that it can only pursue goals and activities that are beneficial to humans. Traditional approaches to value alignment use imitation learning or preference learning to infer the values of humans by observing their behavior. We introduce a complementary technique in which a value-aligned prior is learned from naturally occurring stories which encode societal norms. Training data is sourced from the children's educational comic strip, Goofus & Gallant. In this work, we train multiple machine learning models to classify natural language descriptions of situations found in the comic strip as normative or non-normative by identifying if they align with the main characters' behavior. We also report the models' performance when transferring to two unrelated tasks with little to no additional training on the new task.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"3 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86092689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The implementation of AI systems has led to new forms of harm in various sensitive social domains. We analyze these as problems How to address these harms remains at the center of controversial debate. In this paper, we discuss the inherent normative uncertainty and political debates surrounding the safety of AI systems.of vagueness to illustrate the shortcomings of current technical approaches in the AI Safety literature, crystallized in three dilemmas that remain in the design, training and deployment of AI systems. We argue that resolving normative uncertainty to render a system 'safe' requires a sociotechnical orientation that combines quantitative and qualitative methods and that assigns design and decision power across affected stakeholders to navigate these dilemmas through distinct channels for dissent. We propose a set of sociotechnical commitments and related virtues to set a bar for declaring an AI system 'human-compatible', implicating broader interdisciplinary design approaches.
{"title":"Hard Choices in Artificial Intelligence: Addressing Normative Uncertainty through Sociotechnical Commitments","authors":"Roel Dobbe, T. Gilbert, Yonatan Dov Mintz","doi":"10.1145/3375627.3375861","DOIUrl":"https://doi.org/10.1145/3375627.3375861","url":null,"abstract":"The implementation of AI systems has led to new forms of harm in various sensitive social domains. We analyze these as problems How to address these harms remains at the center of controversial debate. In this paper, we discuss the inherent normative uncertainty and political debates surrounding the safety of AI systems.of vagueness to illustrate the shortcomings of current technical approaches in the AI Safety literature, crystallized in three dilemmas that remain in the design, training and deployment of AI systems. We argue that resolving normative uncertainty to render a system 'safe' requires a sociotechnical orientation that combines quantitative and qualitative methods and that assigns design and decision power across affected stakeholders to navigate these dilemmas through distinct channels for dissent. We propose a set of sociotechnical commitments and related virtues to set a bar for declaring an AI system 'human-compatible', implicating broader interdisciplinary design approaches.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"43 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88777700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Certainty around the regulatory environment is crucial to facilitate responsible AI innovation and its social acceptance. However, the existing legal liability system is inapt to assign responsibility where a potentially harmful conduct and/or the harm itself are unforeseeable, yet some instantiations of AI and/or the harms they may trigger are not foreseeable in the legal sense. The unpredictability of how courts would handle such cases makes the risks involved in the investment and use of AI incalculable, creating an environment that is not conducive to innovation and may deprive society of some benefits AI could provide. To tackle this problem, we propose to draw insights from financial regulatory best-practices and establish a system of AI guarantee schemes. We envisage the system to form part of the broader market-structuring regulatory framework, with the primary function to provide a readily available, clear, and transparent funding mechanism to compensate claims that are either extremely hard or impossible to realize via conventional litigation. We propose at least partial industry-funding, with funding arrangements depending on whether it would pursue other potential policy goals.
{"title":"The AI Liability Puzzle and a Fund-Based Work-Around","authors":"Olivia J. Erd'elyi, G'abor Erd'elyi","doi":"10.1145/3375627.3375806","DOIUrl":"https://doi.org/10.1145/3375627.3375806","url":null,"abstract":"Certainty around the regulatory environment is crucial to facilitate responsible AI innovation and its social acceptance. However, the existing legal liability system is inapt to assign responsibility where a potentially harmful conduct and/or the harm itself are unforeseeable, yet some instantiations of AI and/or the harms they may trigger are not foreseeable in the legal sense. The unpredictability of how courts would handle such cases makes the risks involved in the investment and use of AI incalculable, creating an environment that is not conducive to innovation and may deprive society of some benefits AI could provide. To tackle this problem, we propose to draw insights from financial regulatory best-practices and establish a system of AI guarantee schemes. We envisage the system to form part of the broader market-structuring regulatory framework, with the primary function to provide a readily available, clear, and transparent funding mechanism to compensate claims that are either extremely hard or impossible to realize via conventional litigation. We propose at least partial industry-funding, with funding arrangements depending on whether it would pursue other potential policy goals.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"54 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80985175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As machine learning black boxes are increasingly being deployed in critical domains such as healthcare and criminal justice, there has been a growing emphasis on developing techniques for explaining these black boxes in a human interpretable manner. There has been recent concern that a high-fidelity explanation of a black box ML model may not accurately reflect the biases in the black box. As a consequence, explanations have the potential to mislead human users into trusting a problematic black box. In this work, we rigorously explore the notion of misleading explanations and how they influence user trust in black box models. Specifically, we propose a novel theoretical framework for understanding and generating misleading explanations, and carry out a user study with domain experts to demonstrate how these explanations can be used to mislead users. Our work is the first to empirically establish how user trust in black box models can be manipulated via misleading explanations.
{"title":"\"How do I fool you?\": Manipulating User Trust via Misleading Black Box Explanations","authors":"Himabindu Lakkaraju, O. Bastani","doi":"10.1145/3375627.3375833","DOIUrl":"https://doi.org/10.1145/3375627.3375833","url":null,"abstract":"As machine learning black boxes are increasingly being deployed in critical domains such as healthcare and criminal justice, there has been a growing emphasis on developing techniques for explaining these black boxes in a human interpretable manner. There has been recent concern that a high-fidelity explanation of a black box ML model may not accurately reflect the biases in the black box. As a consequence, explanations have the potential to mislead human users into trusting a problematic black box. In this work, we rigorously explore the notion of misleading explanations and how they influence user trust in black box models. Specifically, we propose a novel theoretical framework for understanding and generating misleading explanations, and carry out a user study with domain experts to demonstrate how these explanations can be used to mislead users. Our work is the first to empirically establish how user trust in black box models can be manipulated via misleading explanations.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"46 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72766489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
William Cai, Johann D. Gaebler, Nikhil Garg, Sharad Goel
Public and private institutions must often allocate scarce resources under uncertainty. Banks, for example, extend credit to loan applicants based in part on their estimated likelihood of repaying a loan. But when the quality of information differs across candidates (e.g., if some applicants lack traditional credit histories), common lending strategies can lead to disparities across groups. Here we consider a setting in which decision makers---before allocating resources---can choose to spend some of their limited budget further screening select individuals. We present a computationally efficient algorithm for deciding whom to screen that maximizes a standard measure of social welfare. Intuitively, decision makers should screen candidates on the margin, for whom the additional information could plausibly alter the allocation. We formalize this idea by showing the problem can be reduced to solving a series of linear programs. Both on synthetic and real-world datasets, this strategy improves utility, illustrating the value of targeted information acquisition in such decisions. Further, when there is social value for distributing resources to groups for whom we have a priori poor information---like those without credit scores---our approach can substantially improve the allocation of limited assets.
{"title":"Fair Allocation through Selective Information Acquisition","authors":"William Cai, Johann D. Gaebler, Nikhil Garg, Sharad Goel","doi":"10.1145/3375627.3375823","DOIUrl":"https://doi.org/10.1145/3375627.3375823","url":null,"abstract":"Public and private institutions must often allocate scarce resources under uncertainty. Banks, for example, extend credit to loan applicants based in part on their estimated likelihood of repaying a loan. But when the quality of information differs across candidates (e.g., if some applicants lack traditional credit histories), common lending strategies can lead to disparities across groups. Here we consider a setting in which decision makers---before allocating resources---can choose to spend some of their limited budget further screening select individuals. We present a computationally efficient algorithm for deciding whom to screen that maximizes a standard measure of social welfare. Intuitively, decision makers should screen candidates on the margin, for whom the additional information could plausibly alter the allocation. We formalize this idea by showing the problem can be reduced to solving a series of linear programs. Both on synthetic and real-world datasets, this strategy improves utility, illustrating the value of targeted information acquisition in such decisions. Further, when there is social value for distributing resources to groups for whom we have a priori poor information---like those without credit scores---our approach can substantially improve the allocation of limited assets.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86362598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Dylan Slack, Sophie Hilgard, Emily Jia, Sameer Singh, Himabindu Lakkaraju
As machine learning black boxes are increasingly being deployed in domains such as healthcare and criminal justice, there is growing emphasis on building tools and techniques for explaining these black boxes in an interpretable manner. Such explanations are being leveraged by domain experts to diagnose systematic errors and underlying biases of black boxes. In this paper, we demonstrate that post hoc explanations techniques that rely on input perturbations, such as LIME and SHAP, are not reliable. Specifically, we propose a novel scaffolding technique that effectively hides the biases of any given classifier by allowing an adversarial entity to craft an arbitrary desired explanation. Our approach can be used to scaffold any biased classifier in such a way that its predictions on the input data distribution still remain biased, but the post hoc explanations of the scaffolded classifier look innocuous. Using extensive evaluation with multiple real world datasets (including COMPAS), we demonstrate how extremely biased (racist) classifiers crafted by our framework can easily fool popular explanation techniques such as LIME and SHAP into generating innocuous explanations which do not reflect the underlying biases.
{"title":"Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods","authors":"Dylan Slack, Sophie Hilgard, Emily Jia, Sameer Singh, Himabindu Lakkaraju","doi":"10.1145/3375627.3375830","DOIUrl":"https://doi.org/10.1145/3375627.3375830","url":null,"abstract":"As machine learning black boxes are increasingly being deployed in domains such as healthcare and criminal justice, there is growing emphasis on building tools and techniques for explaining these black boxes in an interpretable manner. Such explanations are being leveraged by domain experts to diagnose systematic errors and underlying biases of black boxes. In this paper, we demonstrate that post hoc explanations techniques that rely on input perturbations, such as LIME and SHAP, are not reliable. Specifically, we propose a novel scaffolding technique that effectively hides the biases of any given classifier by allowing an adversarial entity to craft an arbitrary desired explanation. Our approach can be used to scaffold any biased classifier in such a way that its predictions on the input data distribution still remain biased, but the post hoc explanations of the scaffolded classifier look innocuous. Using extensive evaluation with multiple real world datasets (including COMPAS), we demonstrate how extremely biased (racist) classifiers crafted by our framework can easily fool popular explanation techniques such as LIME and SHAP into generating innocuous explanations which do not reflect the underlying biases.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"74 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-11-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77347056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Concerns within the machine learning community and external pressures from regulators over the vulnerabilities of machine learning algorithms have spurred on the fields of explainability, robustness, and fairness. Often, issues in explainability, robustness, and fairness are confined to their specific sub-fields and few tools exist for model developers to use to simultaneously build their modeling pipelines in a transparent, accountable, and fair way. This can lead to a bottleneck on the model developer's side as they must juggle multiple methods to evaluate their algorithms. In this paper, we present a single framework for analyzing the robustness, fairness, and explainability of a classifier. The framework, which is based on the generation of counterfactual explanations through a custom genetic algorithm, is flexible, model-agnostic, and does not require access to model internals. The framework allows the user to calculate robustness and fairness scores for individual models and generate explanations for individual predictions which provide a means for actionable recourse (changes to an input to help get a desired outcome). This is the first time that a unified tool has been developed to address three key issues pertaining towards building a responsible artificial intelligence system.
{"title":"CERTIFAI: A Common Framework to Provide Explanations and Analyse the Fairness and Robustness of Black-box Models","authors":"Shubham Sharma, Jette Henderson, Joydeep Ghosh","doi":"10.1145/3375627.3375812","DOIUrl":"https://doi.org/10.1145/3375627.3375812","url":null,"abstract":"Concerns within the machine learning community and external pressures from regulators over the vulnerabilities of machine learning algorithms have spurred on the fields of explainability, robustness, and fairness. Often, issues in explainability, robustness, and fairness are confined to their specific sub-fields and few tools exist for model developers to use to simultaneously build their modeling pipelines in a transparent, accountable, and fair way. This can lead to a bottleneck on the model developer's side as they must juggle multiple methods to evaluate their algorithms. In this paper, we present a single framework for analyzing the robustness, fairness, and explainability of a classifier. The framework, which is based on the generation of counterfactual explanations through a custom genetic algorithm, is flexible, model-agnostic, and does not require access to model internals. The framework allows the user to calculate robustness and fairness scores for individual models and generate explanations for individual predictions which provide a means for actionable recourse (changes to an input to help get a desired outcome). This is the first time that a unified tool has been developed to address three key issues pertaining towards building a responsible artificial intelligence system.","PeriodicalId":93612,"journal":{"name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75945728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：