ARC-AGI has introduced a third stage of its famous benchmark. You can review it here. ARC-AGI3 distances itself from 1 and 2, developing towards a more genuine test of task acquisition. If you play demos of ARC-AGI3, you will see that they are beginning to mimic traditional environments seen in Reinforcement Learning research. Design Philosophy Easy for Humans, Hard for AI At the core of ARC-AGI benchmark design is the the principle of “Easy for Humans, Hard for […]

Ver mais

This really irks me for some reason but there’s like 10,000 explanations for mHC online while the only instance of someone actually trying to explore mHC in code is a single github repo (props to the repo). I just want to be able to implement it and plug it into existing projects. I don’t need yet another analogy for why a cat won’t fall off a cliff the ground isn’t tipped over. This reminds me of my physics […]

Ver mais

Abstract Encryption has long been the cornerstone of digital privacy. From TLS securing web traffic to end-to-end encryption protecting messages, modern systems rely heavily on cryptography to prevent unauthorized access. However, encryption primarily protects data in transit and at rest – not the existence, lifetime, or reuse of data itself. In practice, encrypted systems still accumulate long-lived state: cookies, tokens, browser storage, logs, backups, and metadata that persist long after their intended use. This article argues that ephemerality […]

Ver mais

arXiv:2602.04904v1 Announce Type: new Abstract: Multimodal fusion faces two robustness challenges: noisy inputs degrade representation quality, and missing modalities cause prediction failures. We propose DCER, a unified framework addressing both challenges through dual-stage compression and energy-based reconstruction. The compression stage operates at two levels: within-modality frequency transforms (wavelet for audio, DCT for video) remove noise while preserving task-relevant patterns, and cross-modality bottleneck tokens force genuine integration rather than modality-specific shortcuts. For missing modalities, energy-based reconstruction recovers representations via […]

Ver mais

We propose a federated learning framework for the calibration of parametric insurance indices under heterogeneous renewable energy production losses. Producers locally model their losses using Tweedie generalized linear models and private data, while a common index is learned through federated optimization without sharing raw observations. The approach accommodates heterogeneity in variance and link functions and directly minimizes a global deviance objective in a distributed setting. We implement and compare FedAvg, FedProx and FedOpt, and benchmark them against an […]

Ver mais

arXiv:2602.16914v1 Announce Type: cross Abstract: Modeling of longitudinal cohort data typically involves complex temporal dependencies between multiple variables. There, the transformer architecture, which has been highly successful in language and vision applications, allows us to account for the fact that the most recently observed time points in an individual’s history may not always be the most important for the immediate future. This is achieved by assigning attention weights to observations of an individual based on a transformation of […]

Ver mais

arXiv:2601.16399v1 Announce Type: new Abstract: We study a structured bi-level optimization problem where the upper-level objective is a smooth function and the lower-level problem is policy optimization in a Markov decision process (MDP). The upper-level decision variable parameterizes the reward of the lower-level MDP, and the upper-level objective depends on the optimal induced policy. Existing methods for bi-level optimization and RL often require second-order information, impose strong regularization at the lower level, or inefficiently use samples through nested-loop […]

Ver mais

arXiv:2602.10714v1 Announce Type: cross Abstract: Preconditioning is a common method applied to modify Markov chain Monte Carlo algorithms with the goal of making them more efficient. In practice it is often extremely effective, even when the preconditioner is learned from the chain. We analyse and compare the finite-time computational costs of schemes which learn a preconditioner based on the target covariance or the expected Hessian of the target potential with that of a corresponding scheme that does not […]

Ver mais