Dr. Arya’s current research focuses on developing neuro-symbolic models that integrate deep neural architectures with symbolic reasoning and probabilistic frameworks; advancing tractable probabilistic modeling to support efficient and scalable inference; creating neural network–based solvers for large-scale combinatorial optimization; and designing methods for explainable AI in structured and hybrid systems, with applications in computer vision, activity recognition, and multimodal learning.
Neural Network Approximators for Marginal MAP Inference AAAI’24-Oral
Learning to Solve the Constrained Most Probable Explanation Task AISTATS’24 UAI TPM’24
A Neural Network Approach for Efficiently Answering Most Probable Explanation Queries in Probabilistic Models NeurIPS’24-Spotlight UAI TPM’24-Best Paper Award
SINE - Scalable MPE Inference for Probabilistic Graphical Models using Advanced Neural Embeddings AISTATS’25
Deep Dependency Networks and Advanced Inference Schemes for Multi-Label Classification AISTATS’24
CaptainCook4D Dataset NeurIPS’24 DMLR’23
Explainable Activity Recognition TiiS’23
Predictive Task Guidance in Augmented Reality Poster at IEEE VR’24
Kernelized Random Vector Functional Link Network IJCNN’20