Himabindu Lakkaraju

Last updated

Himabindu Lakkaraju
Alma mater Indian Institute of Science
Stanford University
Scientific career
Institutions University of Chicago
IBM Research
Microsoft Research
Harvard University
Thesis Human-centric machine learning : enabling machine learning for high-stakes decision-making  (2018)
Doctoral advisor Jure Leskovec

Himabindu "Hima" Lakkaraju is an Indian-American computer scientist who works on machine learning, artificial intelligence, algorithmic bias, and AI accountability. She is currently an Assistant Professor at the Harvard Business School and is also affiliated with the Department of Computer Science at Harvard University. Lakkaraju is known for her work on explainable machine learning. More broadly, her research focuses on developing machine learning models and algorithms that are interpretable, transparent, fair, and reliable. She also investigates the practical and ethical implications of deploying machine learning models in domains involving high-stakes decisions such as healthcare, criminal justice, business, and education. Lakkaraju was named as one of the world's top Innovators Under 35 by both Vanity Fair and the MIT Technology Review.

Contents

She is also known for her efforts to make the field of machine learning more accessible to the general public. Lakkaraju co-founded the Trustworthy ML Initiative (TrustML) to lower entry barriers and promote research on interpretability, fairness, privacy, and robustness of machine learning models. [1] She has also developed several tutorials [2] [3] [4] [5] and a full-fledged course on the topic of explainable machine learning. [6]

Early life and education

Lakkaraju obtained a masters degree in computer science from the Indian Institute of Science in Bangalore. As part of her masters thesis, she worked on probabilistic graphical models and developed semi-supervised topic models which can be used to automatically extract sentiment and concepts from customer reviews. [7] [8] This work was published at the SIAM International Conference on Data Mining, and won the Best Research Paper Award at the conference. [9]

She then spent two years as a research engineer at IBM Research, India in Bangalore before moving to Stanford University to pursue her PhD in computer science. Her doctoral thesis was advised by Jure Leskovec. She also collaborated with Jon Kleinberg, Cynthia Rudin, and Sendhil Mullainathan during her PhD. Her doctoral research focused on developing interpretable and fair machine learning models that can complement human decision making in domains such as healthcare, criminal justice, and education. [10] This work was awarded the Microsoft Research Dissertation Grant [11] and the INFORMS Best Data Mining Paper prize. [12]

During her PhD, Lakkaraju spent a summer working as a research fellow at the Data Science for Social Good program at University of Chicago. As part of this program, she collaborated with Rayid Ghani to develop machine learning models which can identify at-risk students and also prescribe appropriate interventions. This research was leveraged by schools in Montgomery County, Maryland. [13] Lakkaraju also worked as a research intern and visiting researcher at Microsoft Research, Redmond during her PhD. She collaborated with Eric Horvitz at Microsoft Research to develop human-in-the-loop algorithms for identifying blind spots of machine learning models. [14]

Research and career

Lakkaraju's doctoral research focused on developing and evaluating interpretable, transparent, and fair predictive models which can assist human decision makers (e.g., doctors, judges) in domains such as healthcare, criminal justice, and education. [10] As part of her doctoral thesis, she developed algorithms for automatically constructing interpretable rules for classification [15] and other complex decisions which involve trade-offs. [16] Lakkaraju and her co-authors also highlighted the challenges associated with evaluating predictive models in settings with missing counterfactuals and unmeasured confounders, and developed new computational frameworks for addressing these challenges. [17] [18] She co-authored a study which demonstrated that when machine learning models are used to assist in making bail decisions, they can help reduce crime rates by up to 24.8% without exacerbating racial disparities. [18] [19]

Lakkaraju joined Harvard University as a postdoctoral researcher in 2018, and then became an assistant professor at the Harvard Business School and the Department of Computer Science at Harvard University in 2020. [20] [21] Over the past few years, she has done pioneering work in the area of explainable machine learning. She initiated the study of adaptive and interactive post hoc explanations [22] [23] which can be used to explain the behavior of complex machine learning models in a manner that is tailored to user preferences. [24] [25] She and her collaborators also made one of the first attempts at identifying and formalizing the vulnerabilities of popular post hoc explanation methods. [26] They demonstrated how adversaries can game popular explanation methods, and elicit explanations that hide undesirable biases (e.g., racial or gender biases) of the underlying models. Lakkaraju also co-authored a study which demonstrated that domain experts may not always interpret post hoc explanations correctly, and that adversaries could exploit post hoc explanations to manipulate experts into trusting and deploying biased models. [27]

She also worked on improving the reliability of explanation methods. She and her collaborators developed novel theory [28] and methods [29] [30] to analyze and improve the robustness of different classes of post hoc explanation methods by proposing a unified theoretical framework and establishing the first known connections between explainability and adversarial training. Lakkaraju has also made important research contributions to the field of algorithmic recourse. She and her co-authors developed one of the first methods which allows decision makers to vet predictive models thoroughly to ensure that the recourse provided is meaningful and non-discriminatory. [25] Her research has also highlighted critical flaws in several popular approaches in the literature of algorithmic recourse. [31]

Trustworthy ML Initiative (TrustML)

In 2020, Lakkaraju co-founded the Trustworthy ML Initiative (TrustML) to democratize and promote research in the field of trustworthy machine learning which broadly encompasses interpretability, fairness, privacy, and robustness of machine learning models. [1] This initiative aims to enable easy access of fundamental resources to newcomers in the field, provide a platform for early career researchers to showcase their work, and more broadly develop a community of researchers and practitioners working on topics related to trustworthy ML.

Lakkaraju has developed several tutorials [2] [3] [4] [5] and a full-fledged course on explainable machine learning [6] as part of this initiative.

Awards and honors

A course on "Interpretability and Explainability in Machine Learning", 2019

NeurIPS conference tutorial on "Explaining Machine Learning Predictions: State-of-the-art, Challenges, and Opportunities ", 2020

AAAI conference tutorial on "Explaining Machine Learning Predictions: State-of-the-art, Challenges, and Opportunities", 2021

CHIL conference tutorial on "Explainable ML: Understanding the Limits and Pushing the Boundaries", 2021

Selected publications

Related Research Articles

Data mining is the process of extracting and discovering patterns in large data sets involving methods at the intersection of machine learning, statistics, and database systems. Data mining is an interdisciplinary subfield of computer science and statistics with an overall goal of extracting information from a data set and transforming the information into a comprehensible structure for further use. Data mining is the analysis step of the "knowledge discovery in databases" process, or KDD. Aside from the raw analysis step, it also involves database and data management aspects, data pre-processing, model and inference considerations, interestingness metrics, complexity considerations, post-processing of discovered structures, visualization, and online updating.

Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can effectively generalize and thus perform tasks without explicit instructions. Recently, generative artificial neural networks have been able to surpass many previous approaches in performance. While machine learning algorithms have shown remarkable performances on various tasks, they are susceptible to inheriting and amplifying biases present in their training data. This can manifest in skewed representations or unfair treatment of different demographics, such as those based on race, gender, language, and cultural groups.

<span class="mw-page-title-main">Learning classifier system</span> Paradigm of rule-based machine learning methods

Learning classifier systems, or LCS, are a paradigm of rule-based machine learning methods that combine a discovery component with a learning component. Learning classifier systems seek to identify a set of context-dependent rules that collectively store and apply knowledge in a piecewise manner in order to make predictions. This approach allows complex solution spaces to be broken up into smaller, simpler parts.

Data Stream Mining is the process of extracting knowledge structures from continuous, rapid data records. A data stream is an ordered sequence of instances that in many applications of data stream mining can be read only once or a small number of times using limited computing and storage capabilities.

<span class="mw-page-title-main">Orange (software)</span>

Orange is an open-source data visualization, machine learning and data mining toolkit. It features a visual programming front-end for explorative qualitative data analysis and interactive data visualization.

Vasant G. Honavar is an Indian-American computer scientist, and artificial intelligence, machine learning, big data, data science, causal inference, knowledge representation, bioinformatics and health informatics researcher and professor.

Rayid Ghani is a Distinguished Career Professor in the Machine Learning Department and the Heinz College of Information Systems and Public Policy at Carnegie Mellon University. Previously, he was the director of the Center for Data Science and Public Policy, research associate professor in the department of computer science, and a senior fellow at the Harris School of Public Policy at the University of Chicago. He was also the co-founder of Edgeflip, an analytics startup that grew out of the Obama 2012 Campaign, focused on social media products for non-profits, advocacy groups, and charities. Recently, it was announced that he will be leaving the University of Chicago and joining Carnegie Mellon University's School of Computer Science and Heinz College of Information Systems and Public Policy.

Jens Otto Ludwig is a University of Chicago economist whose research focuses on social policy, particularly urban issues such as poverty, crime, and education. He is McCormick Foundation Professor of Social Service Administration, Law, and Public Policy in the School of Social Service Administration and Harris School of Public Policy Studies at the University of Chicago, where he also serves as Co-Director of the university's Urban Education and Crime Labs.

<span class="mw-page-title-main">Jure Leskovec</span> Slovene computer scientist

Jure Leskovec is a Slovenian computer scientist, entrepreneur and associate professor of Computer Science at Stanford University focusing on networks. He was the chief scientist at Pinterest.

<span class="mw-page-title-main">Apache SINGA</span> Open-source machine learning library

Apache SINGA is an Apache top-level project for developing an open source machine learning library. It provides a flexible architecture for scalable distributed training, is extensible to run over a wide range of hardware, and has a focus on health-care applications.

<span class="mw-page-title-main">Explainable artificial intelligence</span> AI in which the results of the solution can be understood by humans

Explainable AI (XAI), often overlapping with Interpretable AI, or Explainable Machine Learning (XML), either refers to an AI system over which it is possible for humans to retain intellectual oversight, or to the methods to achieve this. The main focus is usually on the reasoning behind the decisions or predictions made by the AI which are made more understandable and transparent. XAI counters the "black box" tendency of machine learning, where even the AI's designers cannot explain why it arrived at a specific decision.

In the regulation of algorithms, particularly artificial intelligence and its subfield of machine learning, a right to explanation is a right to be given an explanation for an output of the algorithm. Such rights primarily refer to individual rights to be given an explanation for decisions that significantly affect an individual, particularly legally or financially. For example, a person who applies for a loan and is denied may ask for an explanation, which could be "Credit bureau X reports that you declared bankruptcy last year; this is the main factor in considering you too likely to default, and thus we will not give you the loan you applied for."

<span class="mw-page-title-main">ML.NET</span> Machine learning library

ML.NET is a free software machine learning library for the C# and F# programming languages. It also supports Python models when used together with NimbusML. The preview release of ML.NET included transforms for feature engineering like n-gram creation, and learners to handle binary classification, multi-class classification, and regression tasks. Additional ML tasks like anomaly detection and recommendation systems have since been added, and other approaches like deep learning will be included in future versions.

<span class="mw-page-title-main">Cynthia Rudin</span> American computer scientist and statistician

Cynthia Diane Rudin is an American computer scientist and statistician specializing in machine learning and known for her work in interpretable machine learning. She is the director of the Interpretable Machine Learning Lab at Duke University, where she is a professor of computer science, electrical and computer engineering, statistical science, and biostatistics and bioinformatics. In 2022, she won the Squirrel AI Award for Artificial Intelligence for the Benefit of Humanity from the Association for the Advancement of Artificial Intelligence (AAAI) for her work on the importance of transparency for AI systems in high-risk domains.

Fairness in machine learning refers to the various attempts at correcting algorithmic bias in automated decision processes based on machine learning models. Decisions made by computers after a machine-learning process may be considered unfair if they were based on variables considered sensitive. Examples of these kinds of variable include gender, ethnicity, sexual orientation, disability, language, and more. As it is the case with many ethical concepts, definitions of fairness and bias are always controversial. In general, fairness and bias are considered relevant when the decision process impacts people's lives. In machine learning, the problem of algorithmic bias is well known and well studied. Outcomes may be skewed by a range of factors and thus might be considered unfair with respect to certain groups or individuals. An example would be the way social media sites deliver personalized news to consumers.

node2vec is an algorithm to generate vector representations of nodes on a graph. The node2vec framework learns low-dimensional representations for nodes in a graph through the use of random walks through a graph starting at a target node. It is useful for a variety of machine learning applications. Besides reducing the engineering effort, representations learned by the algorithm lead to greater predictive power. node2vec follows the intuition that random walks through a graph can be treated like sentences in a corpus. Each node in a graph is treated like an individual word, and a random walk is treated as a sentence. By feeding these "sentences" into a skip-gram, or by using the continuous bag of words model paths found by random walks can be treated as sentences, and traditional data-mining techniques for documents can be used. The algorithm generalizes prior work which is based on rigid notions of network neighborhoods, and argues that the added flexibility in exploring neighborhoods is the key to learning richer representations of nodes in graphs. The algorithm is considered one of the best graph classifiers.

Human-AI collaboration is the study of how humans and artificial intelligence (AI) agents work together to accomplish a shared goal. AI systems can aid humans in everything from decision making tasks to art creation. Examples of collaboration include medical decision making aids., hate speech detection, and music generation. As AI systems are able to tackle more complex tasks, studies are exploring how different models and explanation techniques can improve human-AI collaboration.

<span class="mw-page-title-main">Margaret Mitchell (scientist)</span> U.S. computer scientist

Margaret Mitchell is a computer scientist who works on algorithmic bias and fairness in machine learning. She is most well known for her work on automatically removing undesired biases concerning demographic groups from machine learning models, as well as more transparent reporting of their intended use.

Automated decision-making (ADM) involves the use of data, machines and algorithms to make decisions in a range of contexts, including public administration, business, health, education, law, employment, transport, media and entertainment, with varying degrees of human oversight or intervention. ADM involves large-scale data from a range of sources, such as databases, text, social media, sensors, images or speech, that is processed using various technologies including computer software, algorithms, machine learning, natural language processing, artificial intelligence, augmented intelligence and robotics. The increasing use of automated decision-making systems (ADMS) across a range of contexts presents many benefits and challenges to human society requiring consideration of the technical, legal, ethical, societal, educational, economic and health consequences.

References

  1. 1 2 "Trustworthy ML". www.trustworthyml.org. Retrieved April 26, 2021.
  2. 1 2 "NeurIPS 2020 Tutorial on Explainable ML".
  3. 1 2 "AAAI 2021 Tutorial on Explainable ML".
  4. 1 2 "FAccT 2021 Tutorial on Explainable ML in the Wild".
  5. 1 2 "CHIL Conference 2021Tutorial on Limits of Explainable ML".
  6. 1 2 "A Course on Interpretability and Explainability in ML".
  7. ↑ Lakkaraju, Himabindu; Bhattacharyya, Chiranjib; Bhattacharya, Indrajit; Merugu, Srujana (2011). "Exploiting Coherence for the Simultaneous Discovery of Latent Facets and associated Sentiments" (PDF). Proceedings of the 2011 SIAM International Conference on Data Mining (PDF). pp. 498–509. doi:10.1137/1.9781611972818.43. ISBN   978-0-89871-992-5. S2CID   632401.
  8. ↑ "Indian Institute of Science" . Retrieved April 26, 2021.
  9. ↑ "SIAM SDM Best Paper Award".
  10. 1 2 "Human-Centric Machine Learning: Enabling Machine Learning for High-Stakes Decision-Making | Computer Science". cse.ucsd.edu. Retrieved April 26, 2021.
  11. 1 2 "Microsoft Research Dissertation Grant Winners". Microsoft . June 27, 2017.
  12. 1 2 "Curriculum Vitae, Lakkaraju" (PDF).
  13. ↑ "Himabindu Lakkaraju". MIT Technology Review. Retrieved April 26, 2021.
  14. ↑ Lakkaraju, Himabindu; Kamar, Ece; Caruana, Rich; Horvitz, Eric (2016). "Identifying unknown unknowns in the open world: representations and policies for guided exploration". AAAI Conference on Artificial Intelligence: 2124–2132. arXiv: 1610.09064 .
  15. ↑ Lakkaraju, Himabindu; Bach, Stephen H.; Leskovec, Jure (August 1, 2016). "Interpretable Decision Sets: A Joint Framework for Description and Prediction". Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Vol. 2016. pp. 1675–1684. doi:10.1145/2939672.2939874. ISBN   9781450342322. PMC   5108651 . PMID   27853627.
  16. ↑ "Learning Cost-Effective and Interpretable Treatment Regimes" (PDF). International Conference on Artificial Intelligence and Statistics (AISTATS).
  17. ↑ Lakkaraju, H.; Kleinberg, J.; Leskovec, J.; Ludwig, J.; Mullainathan, S. (2017). "The Selective Labels Problem: Evaluating Algorithmic Predictions in the Presence of Unobservables". Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Vol. 2017. pp. 275–284. doi:10.1145/3097983.3098066. ISBN   9781450348874. PMC   5958915 . PMID   29780658.
  18. 1 2 Kleinberg, Jon Michael; Lakkaraju, Himabindu; Leskovec, Jure; Ludwig, Jens; Mullainathan, Sendhil (February 1, 2017). "Human Decisions and Machine Predictions". The Quarterly Journal of Economics. National Bureau of Economic Research Working Paper Series. 133 (1): 237–293. doi: 10.3386/W23180 . PMC   5947971 . PMID   29755141.
  19. ↑ "Research Statement – Lakkaraju" (PDF).
  20. ↑ "Himabindu Lakkaraju; ASSISTANT PROFESSOR OF BUSINESS ADMINISTRATION". Harvard Business School. Retrieved April 26, 2021.
  21. ↑ "SEAS Harvard – Lakkaraju".
  22. 1 2 "National Science Foundation, in collaboration with Amazon, awards 11 Fairness in AI grant projects". Amazon Science. February 10, 2021. Retrieved April 26, 2021.
  23. ↑ "NSF Award Search: Award # 2040989 – FAI: Towards Adaptive and Interactive Post Hoc Explanations". www.nsf.gov. Retrieved April 26, 2021.
  24. ↑ Lakkaraju, Himabindu; Kamar, Ece; Caruana, Rich; Leskovec, Jure (2019). "Faithful and Customizable Explanations of Black Box Models". Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society. Vol. 2020. pp. 131–138. doi: 10.1145/3306618.3314229 . ISBN   9781450363242. S2CID   84839090.
  25. 1 2 Rawal, Kaivalya; Lakkaraju, Himabindu (2020). "Beyond Individualized Recourse: Interpretable and Interactive Summaries of Actionable Recourses" (PDF). Advances in Neural Information Processing Systems. 2020. arXiv: 2009.07165 .
  26. ↑ Slack, Dylan; Hilgard, Sophie; Jia, Emily; Singh, Sameer; Lakkaraju, Himabindu (February 7, 2020). "Fooling LIME and SHAP". Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society. Vol. 2019. pp. 180–186. doi: 10.1145/3375627.3375830 . ISBN   9781450371100. S2CID   211041098.
  27. ↑ Lakkaraju, Himabindu; Bastani, Osbert (February 7, 2020). ""How do I fool you?"". Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society. pp. 79–85. doi: 10.1145/3375627.3375833 . ISBN   9781450371100. S2CID   208077044.
  28. ↑ Agarwal, Sushant; Jabbari, Shahin; Agarwal, Chirag; Upadhyay, Sohini; Zhiwei Steven Wu; Lakkaraju, Himabindu (2021). "Towards the Unification and Robustness of Perturbation and Gradient Based Explanations". International Conference on Machine Learning. 2021. arXiv: 2102.10618 .
  29. ↑ Lakkaraju, Himabindu; Arsov, Nino; Bastani, Osbert (2020). "Robust and Stable Black Box Explanations". International Conference on Machine Learning. 2020. arXiv: 2011.06169 .
  30. ↑ Slack, Dylan; Hilgard, Sophie; Singh, Sameer; Lakkaraju, Himabindu (2020). "How Much Should I Trust You? Modeling Uncertainty of Black Box Explanations". arXiv: 2008.05030 [cs.LG].
  31. ↑ Rawal, Kaivalya; Kamar, Ece; Lakkaraju, Himabindu (2020). "Understanding the Impact of Distribution Shifts on Algorithmic Recourse". arXiv: 2012.11788 [cs.LG].
  32. ↑ "2020 Amazon Research Awards Recipients Announced". April 27, 2021.
  33. ↑ "Meet the Innovators Under 35 – MIT Technology Review". MIT Technology Review Events. Retrieved April 26, 2021.
  34. ↑ Fair, Vanity (October 3, 2019). "The Future Innovators Index 2019". Vanity Fair. Retrieved April 26, 2021.
  35. ↑ axsdeny. "ABOUT | Rising Stars in EECS: 2016" . Retrieved April 26, 2021.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.