Content: Workshop on Hybrid Reasoning and Learning (HRL 2018) @ KR 2018 - Program and Accepted Papers

Workshop on Hybrid Reasoning and Learning (HRL 2018) @ KR 2018 - Program and Accepted Papers

Program

Sunday, October 28

08:45-09:00Welcome
09:00-10:00Invited Talk
Scott Sanner, University of Toronto, Canada.
Symbolic Compilation, Inference, and Decision-making with Hybrid and Deep-learned Models (Abstract)
10:00-10:20Radim Nedbal and Luciano Serafini.
Bayesian Inference for SRL with Hybrid Domains (Paper)
10:20-10:40Till Hofmann and Gerhard Lakemeyer.
A Logic for Specifying Metric Temporal Constraints for Golog Programs (Paper)
10:40-11:00Coffee Break
11:00-11:20Stefanie Speichert and Vaishak Belle.
Learning Probabilistic Logic Programs in Continuous Domains (Paper)
11:20-11:40Stefanie Speichert, Andreas Bueff and Vaishak Belle.
Tractable Querying and Learning in Hybrid Domains via Sum-Product Networks (Paper)
11:40-12:00Nandini Ramanan, Gautam Kunapuli, Tushar Khot, Bahare Fatemi, Seyed Mehran Kazemi, David Poole, Kristian Kersting and Sriraam Natarajan.
Structure Learning for Relational Logistic Regression:An Ensemble Approach (Paper)
12:00-12:20Joohyung Lee and Yi Wang.
Weight Learning in a Probabilistic Extension of Answer Set Programs (Paper)
12:20-13:45Lunch
13:45-15:45Joint Session with "Reasoning about Actions and Processes: Highlights of Recent Advances"
13:45-14:45Invited Talk
Guy Van den Broeck, UCLA, USA.
Probabilistic and Logistic Circuits: A New Synthesis of Logic and Machine Learning (Abstract)
14:45-15:00Fredrik Heintz.
Three Examples of Hybrid Reasoning in the Context of Stream Reasoning (Paper)
15:00-15:15Patrick Koopmann and Benjamin ZarrieƟ.
On the Complexity of Verifying Timed Golog Programs over Description Logic Actions (Paper)
15:15-15:30Joohyung Lee.
Recent Developments in Action Languages Based on Extensions of Answer Set Programs
15:30-15:45Paolo Traverso and Luciano Serafini.
Where am I? Let me learn from the real world!
15:45-16:00Coffee Break
16:00-16:20Federico Cerutti and Matthias Thimm.
Probabilistic Augmentations for Knowledge Representation Formalisms (Paper)
16:20-16:40Steffen Schieweck, Gabriele Kern-Isberner and Michael ten Hompel.
Answer Set Programming for a Large, Knowledge-intense Domain: Practical Support of Warehouse Design (Paper)
16:40-17:00Jonas Philipp Haldimann, Marco Wilhelm and Gabriele Kern-Isberner.
Evaluating Reactive ASP by Formal Belief Revision (Paper)
17:00-17:30Discussion

Invited Talks

Symbolic Compilation, Inference, and Decision-making with Hybrid and Deep-learned Models

Scott Sanner, University of Toronto, Canada

Abstract:
There are various sources of hybrid models ranging from probabilistic programs to (partially) deep-learned models that naturally occur in practical applications. While these models can be extremely complex (e.g., mixed discrete and piecewise nonlinear continuous), they still retain significant structure that can be exploited by symbolic compilation, inference, and decision-making methods. In this talk, I will discuss a variety of methods and insights from my own work in this area as well as some exciting directions for future work.

Probabilistic and Logistic Circuits: A New Synthesis of Logic and Machine Learning

Guy Van den Broeck, UCLA, USA

Abstract:
This talk will discuss the role of logical reasoning in statistical machine learning. Generalizations of tractable logical circuits have recently been brought to bear on a variety of machine learning tasks: tractable density estimation, learning distributions subject to logical constraints, deep structured output prediction, supervised and semi-supervised image classification, feature selection, and approximate graphical model inference, achieving (near-)state-of-the-art results in each scenario. I will describe two new circuit representations whose elegant properties have enabled these advances: probabilistic circuits that represent distributions, and logistic circuits that represent statistical classifiers.

Back to workshop page