"COPs, Bandits, and AI for Good"

In recent years, there has been an increasing interest in applying techniques from AI to tackle societal and environmental challenges, ranging from climate change and natural disasters, to food safety and disease spread. These efforts are typically known under the name AI for Good. While many research works in this area have been focusing on designing machine learning algorithms to learn new insights and predict future events from previously collected data, there is another domain where AI has been found to be useful, namely resource allocation and decision making. In particular, a key step in addressing societal and environmental challenges is to efficiently allocate a set of sparse resources to mitigate the problem(s).

"Statistical Methods for Biomedical Informatics"

Increasingly, researchers are turning to statistics and machine learning methods to help improve clinical outcomes and make sense of complex data. Yet traditional statistical methods are often ill-equipped to handle these new settings. In this talk, I will discuss three common problems in biomedical informatics: (1) data privacy (2) sequence counting (e.g., microbiome sequence or RNA-seq) and (3) clinical decision support tools. With respect to data privacy, I will discuss the current state of the art for statistical data privacy and several methods I developed that aim to balance the competing concern between scientific advancement and safeguarding of personal information. With respect to sequence counting, I will present recent work using Bayesian partially identified models to overcome compositional limitations that are common to many other methods in the literature. With respect to clinical decision support tools, I will discuss recent work using pool-based active learning to create personalized bacteremia risk models using partially-labeled data as bacteremia (bacteria present in the blood) has an imperfect diagnosis process due to possible contaminants.

"Responsible AI: Thinking Beyond Data and Models"

The last decade has seen tremendous growth in artificial intelligence (AI) capabilities and its wide-spread adoption in society. Given the impact they have on our social lives, there have also been research on the fairness, accountability, and ethical values that underlie these technologies. While this line of research has gotten great attention in recent years, a majority of this work focuses primarily on mathematical interventions on the often opaque algorithms or models and/or their immediate inputs (data) and outputs (predictions). Such oversimplified mathematical interventions abstract away the underlying societal context where models are conceived, developed, and ultimately deployed. In this talk, I will discuss two strands of my recent work attempting to look beyond the data and models. First, I will discuss a complex systems theory based approach towards modeling societal context that accounts for its dynamic nature, including delayed impacts and feedback loops, and how to bring the expertise from marginalized communities into that process. Second, I discuss how the current literature on algorithmic fairness is rooted in Western concerns, histories, and values, and how this limits its portability to other geographies and cultures, especially in the Global South. In particular, I will discuss our recent work on re-imagining algorithmic fairness for the Indian context.

"Statistical Methods for the Analysis of Sequence Count Data"

Justin Silverman is an Assistant Professor in the College of Information Science and Technology at Penn State University. Justin completed both an M.D. and Ph.D. at Duke University. His Ph.D. work was done under the mentorship of Dr. Lawrence David and Dr. Sayan Mukherjee. Justin’s dissertation work focused on longitudinal modeling and experimental design of host-associated microbiota surveys.

"Digital Libraries and Research Data Management"

Research data is the bedrock of scientific research. From the same data, multiple scientists can draw different conclusions. Sharing research data is thus of paramount importance. Additionally, if we can share and reuse research data, we can perform comparative studies over data obtained by different research projects, design different applications of the data, and extract maximum benefits for the cost incurred to obtain the data. Currently, although there are repositories where scientists can store and host their data, we often face difficulties related to expenses, ease of use, lack of accepted data formats, and metadata standards, lack of individual rewards for sharing of the data, etc. eScience and digital libraries have attempted to address some of these difficulties. In this talk, I will outline some of the issues and solutions related to research data management especially using case studies from the ChemXSeer, ArchSeer, and CiteSeerX projects. I will highlight issues related to data storage, management, retrieval, and the diversity of the data, need for interoperation, preservation and archival, usability and user access, need for standards, security and trustworthiness of the repositories, etc. outlining the progress in each of these areas briefly. I will conclude by summarizing both the success and the open problems in the area.

“Just, Equitable, and Efficient Algorithmic Allocation of Scarce Societal Resources”

Demand for resources that are collectively controlled or regulated by society, like social services or organs for transplantation, typically far outstrips supply. How should these scarce resources be allocated? Any approach to this question requires insights from computer science, economics, and beyond; we must define objectives (foregrounding equity and distributive justice in addition to efficiency), predict outcomes (taking causal considerations into account), and optimize allocations, while carefully considering agent preferences and incentives. Motivated by the real-world problem of provision of services to homeless households, I will discuss our approach to thinking through how algorithmic approaches and computational thinking can help.

“AI for Population Health”

As exemplified by the COVID-19 pandemic, our health and wellbeing depend on a difficult-to-measure web of societal factors and individual behaviors. AI can help us untangle this web and optimize interventions to improve health at a population level, especially for marginalized groups. However, population health applications raise new computational challenges, requiring us to make sense of limited data and optimize decisions under the resulting uncertainty. This talk presents methodological developments in machine learning, optimization, and social networks which are motivated by on-the-ground collaborations on HIV prevention, tuberculosis treatment, and the COVID-19 response. These projects have produced deployed applications and policy impact. For example, I will present the development of an AI-augmented intervention for HIV prevention among homeless youth. This system was deployed and evaluated in a field test enrolling over 700 youth and found to significantly reduce the prevalence of key risk behaviors for HIV.

"Exposure to News in the Digital Age"

Join Sandra González-Bailón, associate professor in the Annenberg School for Communication at the University of Pennsylvania, for her talk, where she will discuss the implications of the divide between informed citizens and news avoiders, and the need to measure exposure across media channels to identify populations that are most likely to be vulnerable to misinformation campaigns.

“Doing Good with Data: Fairly and Equitably”

Can AI, ML and Data Science help help prevent children from getting lead poisoning? Can it help reduce police violence and misconduct? Can it improve vaccination rates? Can it help cities better prioritize limited resources to improve lives of citizens and achieve equity? We’re all aware of the potential of ML and AI but turning this potential into tangible social impact, and more importantly equitable social impact, takes cross-disciplinary training, new methods, and collaborations with governments and non profits. I’ll discuss lessons learned from working on 50+ projects over the past few years with non-profits and governments on high-impact public policy and social challenges in criminal justice, public health, education, economic development, public safety, workforce training, and urban infrastructure. I’ll highlight opportunities as well as challenges around explainability and bias/fairness that need to tackled in order to have social and policy impact in a fair and equitable manner.

“AI for Public Health and Conservation: Learning and Planning in the Data-to-Deployment Pipeline”

With the maturing of AI and multiagent systems research, we have a tremendous opportunity to direct these advances towards addressing complex societal problems. We focus on the problems of public health and wildlife conservation, and present research advances in multiagent systems to address one key cross-cutting challenge: how to effectively deploy our limited intervention resources in these problem domains. We present our deployments from around the world as well as lessons learned that we hope are of use to researchers who are interested in AI for Social Impact. Achieving social impact in these domains often requires methodological advances; we will highlight key research advances in topics such as computational game theory, multi-armed bandits and influence maximization in social networks for addressing challenges in public health and conservation. In pushing this research agenda, we believe AI can indeed play an important role in fighting social injustice and improving society.

CSRAI will be accepting proposals for its inaugural seed funding program through Oct. 15 with projects expected to start in January 2021 and last for up to two years.