Osonde A. Osoba

@OsondeOsoba on Twitter

Codirector, Center for Scalable Computing and Analysis; Senior Information Scientist; Professor, Pardee RAND Graduate School

Santa Monica Office

Education

Ph.D. & M.S. in electrical engineering, University of Southern California; B.S. in electrical and computer engineering, University of Rochester

Media Resources

This researcher is available for interviews.

To arrange an interview, contact the RAND Office of Media Relations at (310) 451-6913, or email media@rand.org.

More Experts

Overview

Osonde Osoba (pronounced “oh-shOwn-day aw-shAw-bah”) is codirector of the Center for Scalable Computing and Analysis (SCAN), a senior information scientist at the RAND Corporation and a professor at the Pardee RAND Graduate School. He has a background in the design and optimization of machine learning algorithms. He has applied his machine learning (ML) expertise to diverse policy areas such as health, defense, and technology policy.

Osoba's recent applied work includes ML modeling for strategic decision-support, reinforcement learning applied to policy-relevant agent-based models, and causal modeling of social & behavioral phenomena. His recent work on the implications of AI/ML has focused on issues of data privacy and fairness in artificial intelligence and algorithmic decision systems generally.

Prior to joining RAND, he was a researcher at the University of Southern California (USC). His research there focused on improving the speed and robustness of popular statistical algorithms like the expectation-maximization (EM) and backpropagation algorithms used in applications like automatic speech recognition. He also made contributions on the robustness and accuracy of approximate Bayesian inference schemes. Osoba received his Ph.D. in electrical engineering from USC and his B.S. in electrical and computer engineering from the University of Rochester.

Research Focus

Pardee RAND Graduate School Courses

Selected Publications

Osonde A. Osoba, Bart Kosko, "Fuzzy Cognitive Maps of Public Support for Insurgency and Terrorism," Journal of Defense Modeling and Simulation, 14(1), 2017

K. Audhkhasi, O. Osoba, and B. Kosko, "Noise-Enhanced Convolutional Neural Networks," Neural Networks, 78, 2016

Osonde Osoba, Bart Kosko, "Noise-enhanced clustering and competitive learning algorithms," Neural Networks, 37, 2013

Osonde Osoba, Sanya Mitaim, and Bart Kosko., "The noisy expectation–maximization algorithm," Fluctuation and Noise Letters, 12(03), 2013

Recent Media Appearances

Interviews: KPBS-TV Online; Newslaundry

Commentary

Bans on Facial Recognition Are Naive. Hold Law Enforcement Accountable for Its Abuse
Broader police reform may be difficult to achieve. But in the long run, it will be more effective than any specific technology ban.
Jun 17, 2020 The Hill
Don't Make the Pandemic Worse with Poor Data Analysis
The need for immediate answers in the face of severe public health and economic distress may create a temptation to relax statistical standards. But urgency should not preclude expert analysis and honest assessments of uncertainty. Mistaken assumptions could lead to counterproductive actions.
May 6, 2020 The RAND Blog
Did No One Audit the Apple Card Algorithm?
Complex, opaque technologies like artificial intelligence provide significant benefits to society. But those benefits don't eliminate the need for accountability and transparency.
Nov 21, 2019 The RAND Blog
Keeping Artificial Intelligence Accountable to Humans
Artificial intelligence (AI) systems are often only as intelligent and fair as the data used to train them. To enable AI that frees humans from bias instead of reinforcing it, experts and regulators must think more deeply not only about what AI can do, but what it should do—and then teach it how.
Aug 20, 2018 TechCrunch
The Human Side of Artificial Intelligence: Q&A with Osonde Osoba
Osonde Osoba has been exploring AI since age 15. He says it's less about the intelligence and more about being able to capture how humans think. He is developing AI to improve planning and is also studying fairness in algorithmic decisionmaking in insurance pricing and criminal justice.
May 1, 2018
Rethinking Data Privacy
Society benefits from the exchange of large-scale data in many ways. Anonymization is the usual mechanism for addressing the privacy of data subjects. Unfortunately, anonymization is broken.
Oct 5, 2016 Inside Sources

Publications

Journal Article
Steps Towards Value-Aligned Systems
This paper argues for a more structured systems-level approach for assessing value-alignment in sociotechnical systems. We highlight how adopting a system perspective improves our ability to explain and address misbehaving algorithms better.
Mar 10, 2021
Journal Article
Technocultural Pluralism: A "Clash of Civilizations" in Technology?
This paper argues that future frictions and conflicts are more likely to stem from the interaction of technology cultures rather than just political or economic frictions.
Mar 10, 2021
Journal Article
Modeling Agent Behaviors for Policy Analysis Via Reinforcement Learning
This paper examines the value of reinforcement learning models as adaptive, high-performing, and behaviorally-valid models of agent decision-making in Agent-based Models.
Mar 9, 2021
Report
Identifying Systemic Bias in the Acquisition of Machine Learning Decision Aids for Law Enforcement Applications
This Perspective describes ways the U.S. Department of Homeland Security can identify and address bias in law enforcement systems that use machine learning and how to prevent such bias at various points in the process of acquiring such systems.
Jan 21, 2021
Report
United States–Japan Research Exchange on Artificial Intelligence: Proceedings from a Pair of Conferences on the Impact of Artificial Intelligence on Work, Health, and Data Privacy and on Disaster Prediction, Resilience, and Recovery
This report captures insights from two conferences that brought together U.S. and Japanese experts on work, health, and data security and on disaster response and disaster modeling to exchange views on artificial intelligence and machine learning.
Dec 7, 2020
Report
Using Natural Language Processing to Code Patient Experience Narratives: Capabilities and Challenges
The authors demonstrate the feasibility of using natural language processing to extract key information from a large set of patient narratives about their health care experiences.
Oct 7, 2020
Draft
Deep Generative Modeling in Network Science with Applications to Public Policy Research
Explores deep generative modeling in network science towards informing key public policy questions.
Sep 30, 2020
Report
Air Dominance Through Machine Learning: A Preliminary Exploration of Artificial Intelligence–Assisted Mission Planning
U.S. air superiority is being challenged by global competitors. In this report, the authors prototype a new artificial intelligence system to help develop and evaluate concepts of operations for the air domain.
May 29, 2020
Draft
A Complex Systems Agenda for Teaching and Conducting Policy Studies
Suggests changes in the teaching and conduct of policy analysis to account for social problems that often occur in complex adaptive systems.
May 27, 2020
Research Brief
Systemic Risk: It's Not Just in the Financial Sector
After a decade of focus on financial firms, new analysis using an innovative approach and new dataset shows that firms in a diverse range of sectors, such as technology and telecommunications, pose a systemic risk to the economy at large.
Feb 3, 2020
Report
Deterrence in the Age of Thinking Machines
This exploratory report provides an initial examination of how artificial intelligence and autonomous systems could affect deterrence and escalation in conventional crises and conflicts.
Jan 27, 2020
Report
Systemic Risk in the Broad Economy: Interfirm Networks and Shocks in the U.S. Economy
Following the 2008 financial crisis, little attention was paid to systemic risks in the economy at large. The authors of this report address that gap, finding evidence that systemic risk extends beyond the financial sector.
Jan 21, 2020
Report
Algorithmic Equity: A Framework for Social Applications
This report is an examination of pathologies in social institutions' use of algorithmic decisionmaking processes. The primary focus is understanding how to evaluate the equitable use of algorithms across a range of specific applications.
Jul 11, 2019
Journal Article
An Artificial Intelligence/Machine Learning Perspective on Social Simulation: New Data and New Challenges
We review issues of data infrastructure and artificial intelligence for social and behavioral modeling. Among the newer machine learning methods,adversarial training and fuzzy cognitive maps have particular unrealized potential.
Jun 11, 2019
Journal Article
Causal Modeling with Feedback Fuzzy Cognitive Maps
Fuzzy cognitive maps model feedback causal relations. We describe FCMs, combined FCMs from multiple experts, and relationships to other dynamic methods. We illustrate the for two social-scientific problems.
Jun 11, 2019
Journal Article
Ethical and Privacy Issues in Social-Behavioral Research
We discuss key ethical and privacy issues: meaningful user notice,assuring usable and accurate access control, anonymizing data, validating algorithms to avoid harms, addressing harms, and deterring abuse.
Jun 11, 2019
Report
Priority Challenges for Social and Behavioral Research and Its Modeling
Social-behavioral modeling is famously hard. This report examines shortcomings and obstacles -- some inherent to complex adaptive systems broadly and some due to current methods and practices -- then discusses steps that deserve priority attention.
Apr 16, 2018
Draft
An Artificial Intelligence/Machine Learning Perspective on Social Simulation: New Data and New Challenges
Reviews the current state of the art in data infrastructure and artificial intelligence approaches that could be valuable for social and behavioral modeling.
Mar 5, 2018
Report
The Risks of Artificial Intelligence to Security and the Future of Work
This Perspective explores the policy implications of the rise of algorithms and artificial intelligence (AI) in two domains of significant importance and public interest: security and employment.
Dec 6, 2017
Report
The Global Landpower Network: Recommendations for Strengthening Army Engagement
The U.S. Army's global landpower network concept integrates, sustains, and advances the Army's efforts to meet U.S. national security guidance emphasizing the importance of working closely with partner nations to achieve U.S. strategic objectives.
Aug 21, 2017
Report
An Intelligence in Our Image: The Risks of Bias and Errors in Artificial Intelligence
Machine learning algorithms and artificial intelligence influence many aspects of life today. This report identifies some of their shortcomings and associated policy risks and examines some approaches for combating these problems.
Apr 5, 2017
Report
Assessing the Need for Supercomputing Resources Within the Pacific Area of Responsibility
This report evaluates the capabilities, challenges, and future options for the Maui High Performance Computing Center, providing important context about the center as decisionmakers work toward resolving its budget gap.
Feb 9, 2017
Draft
Privacy Preservation in the Age of Big Data: A Survey
Surveys the anonymization landscape of approaches for addressing re-identification and identifies the challenges that still must be addressed to ensure the minimization of privacy violations.
Sep 23, 2016
Report
Improving Decision Support for Infectious Disease Prevention and Control: Aligning Models and Other Tools with Policymakers' Needs
This report provides an overview of decision-support tools, including models and nonmodeling approaches, for policymakers and modelers and provides recommendations for using the tools to inform real-world infectious disease-related policy questions.
Jul 12, 2016