J
Jarrid Rector-Brooks
Researcher at University of Michigan
Publications - 10
Citations - 106
Jarrid Rector-Brooks is an academic researcher from University of Michigan. The author has contributed to research in topics: Computer science & Convex function. The author has an hindex of 3, co-authored 3 publications receiving 16 citations.
Papers
More filters
Proceedings ArticleDOI
Biological Sequence Design with GFlowNets
Moksh Jain,Emmanuel Bengio,Alejandro Hernández-García,Jarrid Rector-Brooks,Bonaventure F. P. Dossou,Chanakya Ajit Ekbote,Jie Fu,Micheal Kilgour,Dinghuai Zhang,Lena Simine,Payel Das,Yoshua Bengio +11 more
TL;DR: This work proposes an active learning algorithm leveraging epistemic uncertainty estimation and the recently proposed GFlowNets as a generator of diverse candidate solutions, with the objective to obtain a diverse batch of useful and novel batches with high scoring candidates after each round.
Journal ArticleDOI
Learning GFlowNets from partial episodes for improved convergence and stability
Kanika Madan,Jarrid Rector-Brooks,Maksym Korablyov,Emmanuel Bengio,Moksh Jain,Andrei Cristian Nica,Tom Bosc,Yoshua Bengio,Kolya Malkin +8 more
TL;DR: Sub-trajectory balance as mentioned in this paper is a GFlowNet training objective that can learn from partial action subsequences of varying lengths, which accelerates sampler convergence in previously studied and new environments and enables training GFlowNets in environments with longer action sequences and sparser reward landscapes than what was possible before.
Posted Content
DEUP: Direct Epistemic Uncertainty Prediction
Moksh Jain,Salem Lahlou,Hadi Nekoei,Victor Butoi,Paul Bertin,Jarrid Rector-Brooks,Maksym Korablyov,Yoshua Bengio +7 more
TL;DR: Direct Epistemic Uncertainty Prediction (DEUP) as discussed by the authors is a principled approach for directly estimating epistemic uncertainty by learning to predict generalization error and subtracting an estimate of aleatoric uncertainty, i.e., intrinsic unpredictability.
Journal ArticleDOI
Conditional Flow Matching: Simulation-Free Dynamic Optimal Transport
Alexander Tong,Kolya Malkin,Guillaume Huguet,Yanlei Zhang,Jarrid Rector-Brooks,Kilian Fatras,Guy Wolf,Yoshua Bengio +7 more
TL;DR: Conditional Flow Matching (CFM) as discussed by the authors is a new training objective for continuous normalizing models that does not require the source distribution to be Gaussian or require evaluation of its density.
Journal ArticleDOI
Revisiting Projection-Free Optimization for Strongly Convex Constraint Sets
TL;DR: In this paper, the Frank-Wolfe (FW) optimization under strongly convex constraint sets is revisited and shown to converge to the global optimum with high probability to a stationary point at a rate of O(1/t).