About ACM - AAAI Allen Newell Award

David Blei of Columbia University receives the ACM - AAAI Allen Newell Award.

Blei is recognized for significant contributions to machine learning, information retrieval, and statistics. His signature accomplishment is in the machine learning area of “topic modeling", which he pioneered in the foundational paper “Latent Dirichlet Allocation” (LDA). The applications of topic modelling can be found throughout the social, physical, and biological sciences, in areas such as medicine, finance, political science, commerce, and the digital humanities.

Blei has also been a leader in variational inference (VI), another research area that connects computer science to statistics. VI is an optimization-based methodology for approximate probabilistic inference. Blei’s major contribution to VI has been to develop a novel framework—stochastic variational inference (SVI)—that yielded a quantum leap in the size of problems that can be solved with VI. SVI is in wide use in the AI industry and across the sciences.

Additionally, in his work on discrete choice modelling, Blei has developed a methodology for answering counterfactual queries about changes in prices, which helps to identify complimentary and substitutable pairs of products. This work has built a bridge between computer science and econometrics and has been cited for its impactful use of machine learning modeling.

Bernhard Schölkopf, Max Planck Institute for Intelligent Systems and ETH Zurich, and Stuart J. Russell, University of California at Berkeley, receive the ACM - AAAI Allen Newell Award.

Schölkopf is recognized for his widely used research in machine learning, advancing both mathematical foundations and a broad range of applications in science and industry.

Schölkopf has made fundamental contributions to kernel methods and causality. His contributions to kernel PCA and kernel embeddings have advanced fundamental statistical methodology in dimensionality reduction and hypothesis testing. Professor Schölkopf and his team have advanced numerous areas of applied machine learning, including applications to astronomy, biology, computer vision, robotics, neuroscience, and cognitive science. Schölkopf’s pioneering work in causal machine learning has laid the foundation for a novel understanding of learning causal relationships from data, with implications for all areas of science.

Russell is recognized for a series of foundational contributions to Artificial Intelligence, spanning a wide range of areas such as logical and probabilistic reasoning, knowledge representation, machine learning, reinforcement learning, and the ethics of AI.

Early in his career, Russell defined and studied the concept of bounded optimality, for which he received the 1995 IJCAI Computers and Thought Award. His book, Artificial Intelligence: A Modern Approach (co-authored with Peter Norvig), is the preeminent textbook for AI. It has been used for decades to train AI students in more than 1,500 universities all over the world. Russell’s work on BLOG (Bayesian Logic) led to the creation of the NETVISA global seismic monitoring algorithm that has the capability to reliably detect and accurately localize nuclear explosions. In recent years he has also become an influential figure in addressing ethical issues in AI.

Carla Gomes of Cornell University receives the ACM - AAAI Allen Newell Award for establishing and nurturing the field of computational sustainability and for foundational contributions to artificial intelligence..

Gomes is a leader in AI, particularly in reasoning, optimization, and the integration of learning and reasoning. She is the driving force behind the new subfield of computational sustainability, embodying the values of multidisciplinary research and social impact. Her research advances core computer science and AI while establishing rich connections to other disciplines.

Gomes has played a key role in advancing the integration of methods from AI and operations research. With collaborators, she pioneered randomized restarts and algorithm portfolios for combinatorial solvers. This work has had a tremendous practical impact on solvers for satisfiability (SAT), mixed integer programming (MIP), and satisfiability modulo theories (SMT). Gomes discovered and characterized heavy-tailed runtime distributions and backdoor variables in combinatorial search, explaining the large runtime variations of combinatorial solvers. She also introduced XOR-streamlining, a novel strategy for model counting that was a key step to further advances in efficient probabilistic inference.

Inspired by her early work on experiment design for nitrogen management and wildlife-corridor design, Gomes conceived an ambitious vision for computational sustainability: a highly interdisciplinary research area which incorporates computational thinking to solve critical sustainability challenges.

As the lead principal investigator (PI) of two National Science Foundation (NSF) Expeditions Awards, Gomes has grown Computational Sustainability into a robust and vibrant subfield. She has shown that addressing challenges in sustainability often leads to transformative research in computer science, in addition to having a significant practical impact. Gomes and her collaborators developed a framework for computing the high-dimensional Pareto frontier of ecological and socio-economic tradeoffs of hydro dam expansion in the Amazon Rain Forest.

Gomes also pioneered the use of AI in materials discovery. Together with her team, she developed Deep Reasoning Networks, a novel computational paradigm integrating deep learning with constraint reasoning over rich prior knowledge. This framework was used to solve the crystal-structures phase-mapping problem, which led to the discovery of new solar fuel materials for sustainable energy storage.

Hector Levesque and Moshe Vardi receive the ACM - AAAI Allen Newell Award.

Hector Levesque, University of Toronto, is recognized for fundamental contributions to knowledge representation and reasoning, and their broader influence within theoretical computer science, databases, robotics, and the study of Boolean satisfiability.

Levesque is recognized for his outstanding contributions to the broad core of logic-inspired artificial intelligence and the impact they have had across multiple sub-disciplines within computer science. With collaborators, he has made fundamental contributions to cognitive robotics, multi-agent systems, theoretical computer science, and database systems, as well as in philosophy and cognitive psychology. These have inspired applications such as the semantic web and automated verification. He is internationally recognized as one of the deepest and most original thinkers within AI and a researcher who has advanced the flame that AI pioneer Alan Newell lit.

On the representation side, Levesque has worked on the formalization of several concepts pertaining to artificial and natural agents including belief, goals, intentions, ability, and the interaction between knowledge, perception and action.

On the reasoning side, his research has focused on how automated reasoning can be kept computationally tractable, including the use of greedy local search methods. He is recognized for his fundamental contributions to the development of several new fields of research including the fields of description logic, the study of tractability in knowledge representation, the study of intention and teamwork, the hardness of satisfiability problems, and cognitive robotics. Levesque has also made fundamental contributions to the development of the systematic use of beliefs, desires, and intentions in the development of intelligent software, where his formalization of many aspects of intention and teamwork has shaped the entire approach to the use of these terms and the design of intelligent agents.

Moshe Vardi, Rice University, is cited for contributions to the development of logic as a unifying foundational framework and a tool for modeling computational systems.

Vardi has made major contributions to a wide variety of fields, including database theory, program verification, finite-model theory, reasoning about knowledge, and constraint satisfaction. He is perhaps the most influential researcher working at the interface of logic and computer science, building bridges between communities in computer science and beyond. With his collaborators he has made fundamental contributions to major research areas, including: 1) investigation of the logical theory of databases, where his focus on the trade-off between expressiveness and computational complexity laid the foundations for work on integrity constraints, complexity of query evaluation, incomplete information, database updates, and logic programming in databases; 2) the automata-theoretic approach to reactive systems, which laid mathematical foundations for verifying that a program meets its specifications, and 3) reasoning about knowledge through his development of epistemic logic.

In database theory, Vardi developed a theory of general data dependencies, finding axiomatizations and resolving their decision problem; introduced two basic notions of measuring the complexity of algorithms for evaluating queries, data-complexity, and query-complexity, which soon became standard in the field; created a logical theory of data updates; and characterized the expressive power of query languages and related them to complexity classes.

In software and hardware verification, Vardi introduced an automata-theoretic approach to the verification of reactive systems that revolutionized the field, using automata on infinite strings and trees to represent both the system being analyzed and undesirable computations of the system. Vardi’s automata-theoretic approach has played a central role over the last 30 years of research in the field and in the development of verification tools.

In knowledge theory, Vardi developed rigorous foundations for reasoning about the knowledge of multi-agent and distributed systems, a problem of central importance in many disciplines; his co- authored book on the subject is the definitive source for this field

ACM has named Lydia E. Kavraki of Rice University and Daphne Koller of Stanford University and Insitro recipients of the ACM - AAAI Allen Newell Award.

Lydia Kavraki is recognized for pioneering contributions to robotic motion planning, including the invention of randomized motion planning algorithms and probabilistic roadmaps, with applications to bioinformatics and biomedicine.

Kavraki conducted foundational work on physical algorithms and developed efficient high-dimensional search frameworks that impacted robotics (motion planning, hybrid systems, formal methods in robotics, assembly planning, and micro- and flexible manipulation), as well as computational structural biology, translational bioinformatics, and biomedical informatics.

Kavraki has authored more than 240 peer-reviewed publications and is a co-author of the widely used robotics textbook, Principles of Robot Motion. Her seminal paper, “Probabilistic Roadmaps for Path Planning in High Dimensional Configuration Spaces,” (with Svestka, Latombe and Overmars) was the first to establish a probabilistic approach to developing roadmaps for high-dimensional spaces, which has become one of the key techniques for motion planning for complex physical systems.

Kavraki’s contributions go beyond robotics to address problems underlying the functional annotation of proteins, the understanding of metabolic networks, and the investigation of molecular conformations and protein flexibility. She has contributed to problems that involve reasoning about the three-dimensional structure of biomolecules and their ability to interact with other biomolecules primarily for drug design and, more recently, for personalized cancer immunotherapy.

Daphne Koller is recognized for seminal contributions to machine learning and probabilistic models, the application of these techniques to biology and human health, and for contributions to democratizing education.

Koller was a leader in the development and use of graphical models, including learning the model structure as well as its parameters, and pioneered the unification of statistical learning and relational modelling languages. She also developed foundational methods for inference and learning in temporal models. Her textbook (with Nir Friedman), Probabilistic Graphical Models, is the definitive text in this area.

As an early leader in bringing machine learning methods to the life sciences, she developed Module Networks, wherein she and her colleagues harnessed modularity in gene regulatory programs to build an effective model of gene activity. She has developed groundbreaking applications of machine learning to pathology, work that not only demonstrated the ability of machine learning to outperform human pathologists, but also was one of the first to highlight the importance of the stromal tissue in cancer prognosis (now well-recognized).

Koller is also the co-founder and former co-CEO of Coursera, a platform offering free education from top universities to people worldwide. Coursera, now in its eighth year, has touched the lives of over 50 million learners in every country in the world. Koller is currently the founder and CEO of Insitro, a biotech startup that works to discover better medicines through the integration of machine learning and biology at scale.

Henry Kautz was honored for contributions to artificial intelligence and computational social science, including fundamental results on the complexity of inference, planning and media analytics for public health.

Beginning with his doctoral dissertation, Kautz, now a professor at the University of Rochester, has studied how computers can infer the goals and plans of people by studying their behavior. He has made a range of fundamental contributions to theory and practice in knowledge representation and reasoning, planning and plan recognition and computational social science. Kautz was one of the pioneers in analyzing the computational complexity of knowledge representation formalisms. He was also a co-developer of the first randomized local search algorithms for Boolean satisfiability testing, which have found practical application in planning, graphical models, and software verification.

In the area of pervasive computing and social media analytics, his trailblazing projects have included a system to help cognitively disabled people find their way by inferring the transportation destinations of selected groups of people; a project that uncovered the central role of air travel in the spread of diseases by analyzing social media data; and an initiative to improve the efficiency of restaurant health inspections by combining social media reports of food poisoning with location data.

Margaret A. Boden was honored for contributions to the philosophy and historiography of cognitive science and artificial intelligence, particularly in the study of human creativity. For four decades, Boden has been one of the world’s premiere thought leaders on the intersection of artificial intelligence, cognitive science and the humanities. Through insightful analyses, she was often the first to reveal surprising connections among disciplines, which she then outlined in a series of highly influential books. In two classic books—Artificial Intelligence and Natural Man (1977) and The Creative Mind (1990)—she introduced the idea of creative cognition, which inspired a generation of researchers and laid the foundation of computational creativity as a new subfield of artificial intelligence.

Boden’s two-volume Mind as Machine: A History of Cognitive Science (2006) is a comprehensive survey of interdisciplinary work at the intersection of cognitive science and computation that highlights work in neuroscience, philosophy, biology, linguistics and computation. The book is considered a valuable reference for young researchers to explore what important work has already been undertaken in their field. Mind as Machine also begins with the provocative question: “When is a program not a program?” It is a fitting introduction to Boden, who has been raising important questions about the challenges and risks of artificial intelligence since the beginning of her career. After more than five decades in the field, Boden continues to be both prolific and relevant. Her most recent book AI, Its Nature and Future (2016) was described by Nature magazine as “a masterclass.”

Jitendra Malik was honored for seminal contributions to computer vision that have led the field in image segmentation and object category recognition. It is estimated that 50% of brain power is devoted to visual processing. And as vision is the primary way in which we engage with the world, artificial intelligence (AI) systems which seek to mimic human cognition must incorporate computer vision. For example, it is important for an autonomous vehicle to distinguish between a large plastic bag on the roadway and the limb of a tree.

Malik is recognized as one of the world’s leading researchers in computer vision. He and his lab at UC Berkeley solved several important problems in computer vision, including how to remove “noise” from images in order to identify critical elements such as edges, how to segment images, and how to represent and match shapes. Many of Malik’s former graduate students are now recognized leaders in the field. Through his own research and mentorship, Malik has been a driving force in transforming computer vision from a niche interest to a successful and influential discipline. Computer vision plays an increasingly important role in social media, Internet search, entertainment and autonomous vehicle development.