A new error metric is introduced that precisely captures model vulnerability to consistent adversarial attacks -- perturbations that preserve the ground-truth labels, offering theoretical insight into the mechanisms underlying model sensitivity to adversarial attacks.
Matteo Vilucchio, Lenka Zdeborová, Bruno Loureiro· arXiv.org· 1 citation
This paper considers the estimation of quantiles via a smoothed version of the stochastic gradient descent (SGD) algorithm. By smoothing the score function with a bandwidth tied to the learning rate, we obtain estimates that are monotone in the quantile level at every iteration, while retaining the memory and computati...
Li-Kai Chen, Georg Keilbar, Wei-Biao Wu· 2 citations
This work builds on a principled continuous-time approximation of Markov algorithms and introduces a new, exact entropy flow formula for such processes, and establishes novel connections to a well-studied family of modified logarithmic Sobolev inequalities.
Benjamin Dupuis, Maxime Haddouche, George Deligiannidis et al.· 2 citations
A deep learning based approximation algorithm for solutions of stochastic partial differential equations (SPDEs) using a deep neural network for every realization of the driving noise process of a SPDE to approximate the solution process of the SPDE under consideration.
C. Beck, S. Becker, Patrick Cheridito et al.· arXiv.org· 16 citations· ⚡2
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work investigates how the model parameters -- mainly attention heads and head dimensions -- should be allocated across layers to balance expressivity and efficiency, and proposes principled strategies for allocating attention heads and dimensions across Transformers' layers.
Ruoxi Yu, Hao-Tian Jiang, Jingpu Cheng et al.· 1 citation
It is shown decoder-only MDMs, despite a larger modeling space, can achieve significant inference speedups and comparable perplexity with techniques like temperature annealing, offering a path to reduced inference compute.
Shuchen Xue, Tianyu Xie, Tianyang Hu et al.· arXiv.org· 17 citations· ⚡1
A multi-view linear Structural Equation Model (SEM) is proposed that extends the well-known framework of non-Gaussian disturbances by alternatively leveraging correlation over views and proves the identifiability of the model for acyclic SEMs.
Ambroise Heurtebise, Omar Chehab, Pierre Ablin et al.· 0 citations
Quantized ABBA (QABBA), a quantized version of ABBA, provides an error-controlled, low-precision symbolic representation for time-series compression and LLM-based analysis.
Erin Carson, Xinye Chen, Fei He et al.· 1 citation· ⚡1
This paper forms the resulting $\ell_0$-constrained empirical newsvendor problem with $\ell_2$-regularization, establishes its computational hardness, and develops a mixed-integer second-order cone programming reformulation that strengthens the standard Big-$M$ formulation.
This work establishes generative augmentation as a distributional perturbation process whose reliability can be quantified through Wasserstein-based measures and supported by finite-sample generalization guarantees.
Chathurika S. Abeykoon, M. Muia, Mallory Goldstein· 0 citations
This work addresses the inverse problem of matched score queries at noise scales $\sigma$ and $\lambda\sigma$ using matched subtraction to establish a perturbation bound and a conditional kernel-density-estimation rate.
This paper formalizes a mathematical framework for the Artificial Rosetta Stone, proposing the Artificial Rosetta Stone (ARS), and separates three claims often conflated: a symbolic sequence can be reconstructed probabilistically; a sequence can be consistent with an explicit grammar; and a historical performance can b...
Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.
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.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
