Irit Chelly - Consistent Amortized Clustering via Generative Flow Networks (Heb)

Time and Place
Thursday, January 1st, 2026, 10:30 AM, room C221

Speaker
Irit Chelly (BGU)

Title
Consistent Amortized Clustering via Generative Flow Networks

Abstract:
Neural models for amortized probabilistic clustering yield samples of cluster labels given a set-structured input, while avoiding lengthy Markov chain runs and the need for explicit data likelihoods. Existing methods which label each data point sequentially, like the Neural Clustering Process, often lead to cluster assignments highly dependent on the data order. Alternatively, methods that sequentially create full clusters, do not provide assignment probabilities. In this paper, we introduce GFNCP, a novel framework for amortized clustering. GFNCP is formulated as a Generative Flow Network with a shared energy-based parametrization of policy and reward. We show that the flow matching conditions are equivalent to consistency of the clustering posterior under marginalization, which in turn implies order invariance. GFNCP also outperforms existing methods in clustering performance on both synthetic and real-world data.. The talk is based on [Chelly et. all, AISTATS '25].

Link: https://arxiv.org/pdf/2502.19337

Bio:
Irit Chelly is a PhD graduate from the Computer Science Department at Ben-Gurion University, where she also earned her M.Sc., under the supervision of Prof.Oren Freifeld and Dr. Ari Pakman in the Vision, Inference, and Learning group.Her research focuses on probabilistic clustering using non-parametric Bayesian models and unsupervised learning. Her previous projects involved spatial transformations, dimensionality reduction in video analysis, and generative models. Irit won the national-level Aloni PhD Scholarship from Israel’s Ministry of Technology and Science, as well as the BGU Hi-Tech Scholarship for outstanding PhD students, and received annual awards and instructor rank for excellence in teaching core Computer Science courses.