Temporal Difference Learning for Diffusion Models (ICML 2026) arxiv.org/abs/2606.15048 By Yangchen Pan (my former PhD student) and co-authors. It reformulates diffusion training as a Markov reward process and introduces a TD objective to encourage temporal consistency across d…
Temporal Difference Learning for Diffusion Models arxiv.org
AI Weekly's analysis
→
- The paper introduces a temporal difference objective that penalises inconsistency across the full denoising trajectory rather than only at adjacent time steps.
- It reframes diffusion as a Markov reward process and denoising as a policy evaluation problem, unifying discrete-time and continuous-time formulations.
- Reported FID gains are strongest when the number of sampling steps is small, the regime where few-step samplers and low-compute serving live.
Read full analysis →
View on Bluesky ·
♥ 14
↻ 2
↩ 1
·
2 from the directory shared this ·
33d ago
↻
Amir-massoud Farahmand reposted
@continual-rl.bsky.social
📚 Interested in Continual RL but not sure where to start? 👇 Dive in: sites.google.com/view/continu... ♾️ We've curated a resource hub with key papers, benchmarks, codebases, tutorials, and more to help you get up to speed quickly. #ContinualRL #ReinforcementLearning #MachineLe…
Continual RL Workshop - Resources sites.google.com
View on Bluesky →
There are a few more recent work, but as far as I recall, they are still under restrictive assumptions. There is a new paper came last month, but I haven't read it yet. Convergence of Monte Carlo Optimistic Policy Iteration: Beyond Uniform State-Action Updates arxiv.org/abs/26…
Convergence of Monte Carlo Optimistic Policy Iteration: Beyond Uniform State-Action Updates arxiv.org
There is John Tsitsiklis's work from 2002, but that's for the initial state update model, which feels sample inefficient. Moreover, it assumes either synchronous updates (all states are updated at the same time) or asynchronous but random with uniform distribution. www.jmlr.or…
jmlr.org
