Daniel Selsam
YOU?
Author Swipe
View article: Competitive Programming with Large Reasoning Models
Competitive Programming with Large Reasoning Models Open
We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early chec…
View article: Lean Formalization of Completeness Proof for Coalition Logic with Common Knowledge
Lean Formalization of Completeness Proof for Coalition Logic with Common Knowledge Open
Coalition Logic (CL) is a well-known formalism for reasoning about the strategic abilities of groups of agents in multi-agent systems. Coalition Logic with Common Knowledge (CLC) extends CL with operators from epistic logics, and thus with…
View article: $k$-Equivalence Relations and Associated Algorithms
$k$-Equivalence Relations and Associated Algorithms Open
Lines and circles pose significant scalability challenges in synthetic geometry. A line with $n$ points implies ${n \choose 3}$ collinearity atoms, or alternatively, when lines are represented as functions, equality among ${n \choose 2}$ d…
View article: Universal Policies for Software-Defined MDPs
Universal Policies for Software-Defined MDPs Open
We introduce a new programming paradigm called oracle-guided decision programming in which a program specifies a Markov Decision Process (MDP) and the language provides a universal policy. We prototype a new programming language, Dodona, t…
View article: Sealing pointer-based optimizations behind pure functions
Sealing pointer-based optimizations behind pure functions Open
Functional programming languages are particularly well-suited for building automated reasoning systems, since (among other reasons) a logical term is well modeled by an inductive type, traversing a term can be implemented generically as a …
View article: Guiding High-Performance SAT Solvers with Unsat-Core Predictions
Guiding High-Performance SAT Solvers with Unsat-Core Predictions Open
The NeuroSAT neural network architecture was recently introduced for predicting properties of propositional formulae. When trained to predict the satisfiability of toy problems, it was shown to find solutions and unsatisfiable cores on its…
View article: NeuroCore: Guiding CDCL with Unsat-Core Predictions
NeuroCore: Guiding CDCL with Unsat-Core Predictions Open
The NeuroSAT neural network architecture was recently introduced for predicting properties of propositional formulae. When trained to predict the satisfiability of toy problems, it was shown to find solutions and unsatisfiable cores on its…
View article: Learning a SAT Solver from Single-Bit Supervision
Learning a SAT Solver from Single-Bit Supervision Open
We present NeuroSAT, a message passing neural network that learns to solve SAT problems after only being trained as a classifier to predict satisfiability. Although it is not competitive with state-of-the-art SAT solvers, NeuroSAT can solv…
View article: Developing Bug-Free Machine Learning Systems With Formal Mathematics
Developing Bug-Free Machine Learning Systems With Formal Mathematics Open
Noisy data, non-convex objectives, model misspecification, and numerical instability can all cause undesired behaviors in machine learning systems. As a result, detecting actual implementation errors can be extremely difficult. We demonstr…
View article: Congruence Closure in Intensional Type Theory
Congruence Closure in Intensional Type Theory Open
Congruence closure procedures are used extensively in automated reasoning and are a core component of most satisfiability modulo theories solvers. However, no known congruence closure algorithms can support any of the expressive logics bas…
View article: Data Programming: Creating Large Training Sets, Quickly.
Data Programming: Creating Large Training Sets, Quickly. Open
Large labeled training sets are the critical building blocks of supervised learning methods and are key enablers of deep learning techniques. For some applications, creating labeled training sets is the most time-consuming and expensive pa…
View article: Data Programming: Creating Large Training Sets, Quickly
Data Programming: Creating Large Training Sets, Quickly Open
Large labeled training sets are the critical building blocks of supervised learning methods and are key enablers of deep learning techniques. For some applications, creating labeled training sets is the most time-consuming and expensive pa…
View article: Declarative Probabilistic Programming with Datalog
Declarative Probabilistic Programming with Datalog Open
Probabilistic programming languages are used for developing statistical models, and they typically consist of two components: a specification of a stochastic process (the prior), and a specification of observations that restrict the probab…