
conjunction *with the* first

Machine Learning and Nonparametric Statistics Researcher

conjunction *with the* first

🚨 Local dependence → manifold hypothesis Distant independence → Markov random fields New preprint: the latter works in conjunction first, massivly improves data efficiency, and can be exploited by DNNs arxiv.org/abs/2411.15095 With @itsrainingdata sites.google.com/view/wmtai



NeurIPS preprint: arxiv.org/abs/2410.07685 Standard models for spatial, sequential, hierarchal, etc data result in drastically reduced effective dimension for density estimation. This effective dimension is orthogonal to other approaches such as sparsity or manifold hypothesis.

🎉I'm happy to announce that our paper "Breaking the curse of dimensionality in structured density estimation" was accepted to NeurIPS 2024. 🎉 with: Wai Ming Tai (Nanyang Tech. Uni. - sites.google.com/view/wmtai ) and @itsrainingdata (U Chicago) -Preprint forthcoming-

Just noticed that Tsybakov (author of one of my favorite textbooks: Intro. to Nonparametric Estimation) wrote a nice extension of my 2021 NeurIPS paper "Beyond Smoothness: Incorporating Low-Rank Analysis into Nonparametric Density Estimation." Exciting! hal.science/hal-04557030/

Congratulations: The paper “Set Learning for Accurate and Calibrated Models” is accepted at the #ICLR2024. Authors: @lukas_mut, @robvdm, Q. Zhang, T. Unterthiner, K.-R. Müller. Preprint: arxiv.org/abs/2307.02245 Wrap up: t1p.de/ijaiw @GoogleDeepMind @ml_tuberlin

OKO got accepted to #ICLR2024🎉🥳 OKO is very simple to use: Just change your sampling from single data point sampling to set sampling and compute cross-entropy for the sum of set data point logits rather than for single data point logits. You will get much better calibration! 🔥
🚨 Preprint alert🚨 Are you worried about miscalibration of your model outputs but tired of post-hoc temperature scaling and think label smoothing is a hack? Then just train with OKO - a principled set learning approach for better model calibration! (1/5) arxiv.org/abs/2307.02245

Fun work between @ml_tuberlin @bifoldberlin @GoogleDeepMind 🎆 Thanks to my co-authors @robvdm @TomUnterthiner @richard_qiuyi_z and Klaus-Robert Müller!

🔥The CfP for Re-Align is up 🔥 If you are broadly interested in the alignment of two or more representation spaces, consider submitting to our workshop at #ICLR2024 and join us in beautiful Vienna!
🚨Call for Papers🚨 The Re-Align Workshop is coming to #ICLR2024 Our CfP is finally up! Come share your representational alignment work at our interdisciplinary workshop at @iclr_conf: representational-alignment.g… 1/4

Asking a famous researcher in field X about X. youtu.be/ZDNgmdGMpuY?si=axie…

There are postdoc positions available at my research institute!
👨💻Group: Machine Learning Group, 2 positions 👨🏫Lead: Prof. KR Müller 🧠Job: Research Assistant, E13 🔥Deadline: Jan 15, 2024 🔗t1p.de/tjunf #MachineLearning @ml_tuberlin

At the poster for @lukas_mut's paper "Improving neural network representations using human similarity judgments" x.com/AndrewLampinen/status/… Wed 13 Dec 10:45 a.m. CST — 12:45 p.m. CST Great Hall & Hall B1+B2 (level 1) #300 nips.cc/virtual/2023/poster/…
What aspects of human conceptual knowledge do machine learning objectives fail to adequately capture? Could we use small datasets of human similarity judgements to improve model representations and get better downstream performance? @lukas_mut's cool new work has the answers!

It seems like all math tests for LLMs involve just the direct application of various rules, like a bunch of trig or calc rules, to solve the problem. How about getting a LLM to prove the intermediate value theorem or some other proof that requires a clever construction?

Despite Germans constantly warning me not to, I sit on cold rocks all the time. I guess we Americans just have more robust kidneys.