Min-max optimization (used among other applications in GANs, and adversarial training more broadly) is empirically challenging. We show why min-max optimization is hard in the following paper with Stratis Skoulakis and Manolis Zampetakis:.

💪.

RT @EugeneVinitsky: Hiring researchers and engineers for a stealth, applied research company with a focus on RL x foundation models. Folks….

RT @cmcaram: 🚀 🇬🇷 A year in the making! I’ve just completed a set of 21 lectures in Machine Learning, in Greek, designed for high school st….

Proud of my brother's research!.

Looking forward to this event on Thursday!

RT @NikosDaskalakis: 💡Daskalakis & Ressler labs @McLeanHospital @harvardmed are extremely excited & proud 🥳to share that our collaborative….

Can you train a generative model using only noisy data? If you can, this would alleviate the issue of training data memorization plaguing certain genAI models. In exciting work with @giannis_daras and @AlexGDimakis we show how to do this for diffusion-based generative models.

RT @sxysun1: Stoked to announce an Agentic Markets workshop @agenticmarkets at #ICML 2024! @icmlconf. 📇 Details: ✏️….

RT @JTasioulas: The call for papers is coming soon… Delighted to confirm the participation of the Greek Prime Minister @kmitsotakis and the….

Note that concurrent work by Peng & Rubinstein establishes a similar set of results:

Check out our paper at:

At a technical level, our upper bound employs a tree structure on a collection of external regret learners. Each external regret learner updates its action distribution only sporadically, allowing us to avoid polynomial dependence on the number of experts.

The number of rounds in our upper bound depends exponentially on the inverse of the approximation parameter. We present a lower bound showing that this dependence is necessary, even when the adversary is oblivious.

They also provide algorithms for computing low-rank and sparse correlated equilibria.

As a consequence of our main results and the standard connection between no-swap regret learning and correlated equilibria, we obtain a polynomial-time algorithm for finding an approximate correlated equilibrium in extensive-form games, for constant approximation parameter.

More generally, we obtain that any class of finite Littlestone dimension (or seq fat shattering dimension) has a no swap regret learner. We also derive an extension to the bandit setting, for which we show a nearly-optimal swap regret bound for large action spaces.

We show that no-swap regret learning is possible for any function class which has a no-external regret learner. As an example, in the setting of learning with expert advice with N experts, only polylog(N) rounds are needed to obtain swap regret bounded by an arbitrary constant.

Specifically, we ask: can one improve upon classical bounds by Blum-Mansour and Stolz-Lugosi which suffer linear dependence on the #actions to achieve sublinear swap regret/correlated equilibrium computation with poly-logarithmic dependence on the #actions instead?.

Exciting work w/ @YuvalDagan3 @MFishelson @GolowichNoah on efficient algos for no-swap regret learning and, relatedly, correlated eq when the #actions is exponentially large/infinite. While classical works point in the opposite direction, we show that this is actually possible!.

RT @AlexGDimakis: Athens has a lot of talented AI researchers and developers.