It is shown that while observational outcomes can provide valuable supervision, directly fine-tuning models on such data can lead them to learn spurious correlations, and proposes DeconfoundLM, a method that explicitly removes the effect of known confounders from reward signals.
When providing forecasted probabilities with a predictive model, the ideal model offers perfect calibration: the true probability of the outcome (i.e., the probability that $Y=1$) exactly matches the forecasted probability $f(X)$. In practice, models inevitably exhibit calibration error, and it is therefore important t...
In many settings involving stochastic differential equations, including in diffusion based generative AI, our aim is to accurately generate samples from a terminal distribution. Typically, this is done by generating i.i.d. samples of diffusion paths. Given a fixed simulation budget, a reasonable way to gain efficiency...
We study an online sketched Newton method that approximates the Newton direction at each step via a state-of-the-art sketching solver, called the generalized accelerated sketch-and-project solver (GAS), thereby mitigating the computational bottleneck of classical second-order methods. The GAS solver improves upon vanil...
Xinchen Du, Elizaveta Rebrova, Micha{\l} Derezi\'{n}ski et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Pairwise Likelihood MIA (PL-MIA), a unified method that combines a Gaussian likelihood-ratio (GLR) statistic with population calibration and the Cauchy combination test, is proposed, demonstrating how statistical principles can turn noisy model outputs into more powerful, calibrated, and reproducible evidence for membe...
Sheng-Jie Niu, Ze-Bin Yun, Ye-Heng Ge et al.· 0 citations
This work proposes a learning procedure for system identification in interacting particle systems from single-snapshot observations of collective behaviors, unlike existing approaches that rely on observations of trajectories, and test it on a variety of representative models with steady-state and quasi-stationary patt...
Bao-Li Hao, Mauro Maggioni, Ming Zhong· 0 citations
This work studies tree tensor networks (TTNs), a model class that generalizes deep linear networks and Tucker decompositions and shows they embed arbitrary read-once Boolean formulas, and thus contain polynomial-size targets that cannot be learned by gradient descent in polynomial time under the same mechanism as neura...
Zach Furman, Stephan Wäldchen, Yang-Da Bei et al.· 0 citations
Transfer learning explores how to leverage knowledge from various tasks or domains (sources) to enhance predictive performance in related tasks or domains (targets). Typically, transfer learning research is segmented into several isolated sub-areas (such as domain generalisation, domain adaptation, or multi-domain lear...
Ricardo Ribeiro Pereira, Jacopo Bono, H. Ferreira et al.· 0 citations
Every step from Adam's momentum and denominator to the spectral hitting law is formalized inside the closed state-evolution model, and it is proved that a stronger arbitrary-gradient Adam theorem is impossible, and that exact two-step linear-network loss dynamics do not identify factor spectra or heavy-tail hitting tim...
Theoretically, it is proved that exact early loss dynamics in linear networks do not determine factor spectral tails, that Adam recurrences alone do not imply spectral redistribution, and that projected singular-basis spreading implies contraction of a spectral-tail potential and hence a dimension-corrected hitting-tim...
Classical convergence guarantees for stochastic gradient methods typically assume Lipschitz-smooth objectives and finite-variance gradient noise, both frequently violated in practice. In contrast, we study nonconvex stochastic optimization under the joint relaxation of these assumptions: objectives with $(L,s)$-H\"olde...
Low-Rank Adaptation (LoRA) is an effective approach for adapting large pretrained models by learning low-rank weight updates. In practice, the LoRA rank is used to control an adapter's parameter budget and representational capacity. We show that this view is incomplete: while the nominal rank determines the representat...
Zi-Han Zhu, Zhe-Hang Du, Xu-Yang Chen et al.· 0 citations
Learn how Google DeepMind and GovTech Singapore uses lightweight on-device Gemma models to intercept conversational scams in private chats with zero cloud data egress.
MIT News · Artificial Intelligence· news.mit.eduOct 8, 2026
