Country
Appendix for: Barking up the right tree: an approach to search over molecule synthesis DAGs John Bradshaw
The advantage of the DAG formalism comes at generation time.
moleculesynthesisDAGs
Wecanbreakupthese developments into two connected goals: G1.Learning strong generative models of molecules that can beused tosample novelmolecules, fordownstream screening and scoring tasks; and G2.
Generating Behaviorally Diverse Policies with Latent Diffusion Models
Quality Diversity (QD) is an emerging field in which collections of high performing, behaviorally diverse solutions are trained.
2 Background Diffusion models [53] are latent variable models of the formpฮธ(x0): = R
We show that diffusion models actually are capable of generating high quality samples, sometimes better than the published results on other types of generative models (Section 4). In addition, we show that a certain parameterization of diffusion models reveals an equivalence with denoising score matching over multiple noise levels during training and with annealed Langevin dynamics during sampling (Section 3.2) [55, 61].
3ffebb08d23c609875d7177ee769a3e9-Paper.pdf
Networksadjustingnetworks. In Proceedingsof" Distributed Adaptive Neural Information Processing", pages 197-208, 1990.
ForsterDecompositionandLearningHalfspaceswith Noise
Recentwork[DGT19]obtained thefirstcomputationally efficient learning algorithm with non-trivial error guarantee for this problem.
Supplementary: WhatMakesforGoodViewsfor ContrastiveLearning?
For the semi-supervised view learning, we found the training is stable across different learning ratecombinations, which weconsidered asanadvantage.
WhatMakesforGoodViewsforContrastive Learning?
Contrastive learning between multiple views of the data has recently achieved stateoftheartperformance inthefieldofself-supervised representation learning.