arXiv:2601.19007v1 Announce Type: new Abstract: We propose a novel approach to computationally efficient GP training based on the observation that square-exponential (SE) covariance matrices contain several off-diagonal entries extremely close to zero. We construct a principled procedure to eliminate those entries to produce a emph{banded}-matrix approximation to the original covariance, whose inverse and determinant can be computed at a reduced computational cost, thus contributing to an efficient approximation to the likelihood function. We provide a theoretical analysis of […]

Ver mais

arXiv:2604.08829v1 Announce Type: cross Abstract: The Hierarchical Kernel Transformer (HKT) is a multi-scale attention mechanism that processes sequences at L resolution levels via trainable causal downsampling, combining level-specific score matrices through learned convex weights. The total computational cost is bounded by 4/3 times that of standard attention, reaching 1.3125x for L = 3. Four theoretical results are established. (i) The hierarchical score matrix defines a positive semidefinite kernel under a sufficient condition on the symmetrised bilinear form (Proposition […]

Ver mais

From RAG to Agentic Memory: A Bio‑Inspired Approach to Long‑Term Memory in LLM Agents Continue reading on Towards AI »

Ver mais

arXiv:2601.09032v1 Announce Type: new Abstract: The advancement of large language model (LLM) based agents has shifted AI evaluation from single-turn response assessment to multi-step task completion in interactive environments. We present an empirical study evaluating frontier AI models on 150 workplace tasks within a realistic e-commerce RL environment from Surge. Our analysis reveals an empirically-derived emph{hierarchy of agentic capabilities} that models must master for real-world deployment: (1) tool use, (2) planning and goal formation, (3) adaptability, (4) groundedness, […]

Ver mais

Class imbalance induces systematic bias in deep neural networks by imposing a skewed effective class prior. This work introduces the Neural Prior Estimator (NPE), a framework that learns feature-conditioned log-prior estimates from latent representations. NPE employs one or more Prior Estimation Modules trained jointly with the backbone via a one-way logistic loss. Under the Neural Collapse regime, NPE is analytically shown to recover the class log-prior up to an additive constant, providing a theoretically grounded adaptive signal without […]

Ver mais

Google Antigravity is a powerhouse. As an agentic development platform, it’s remarkably capable of remembering context and executing complex instructions. But after using it to build a turn-based algorithmic simulation of Genesis, I realized something critical: The platform is only as smart as the architect directing it. If you aren’t a developer, you’ll find yourself staring at a “perfect” piece of software that doesn’t actually work. Here is how I used my engineering intuition to bridge the gap […]

Ver mais

My problem is such: I have to use the cpu to train my agent , so running 1600 steps per episode on bipedalwalker, half cheetah etc is out of the question. Are 200 steps fine as a starter point ( assuming the agent can get a score 300 for 1600 steps, that would set the score at 37.5 for 200 steps) so if the agent is able to get to 40 score then for testing I could just […]

Ver mais

arXiv:2502.03174v2 Announce Type: replace-cross Abstract: In this paper, we investigate the label shift quantification problem. We propose robust estimators of the label distribution which turn out to coincide with the Maximum Likelihood Estimator. We analyze the theoretical aspects and derive deviation bounds for the proposed method, providing optimal guarantees in the well-specified case, along with notable robustness properties against outliers and contamination. Our results provide theoretical validation for empirical observations on the robustness of Maximum Likelihood Label Shift.

Ver mais

Learning a stationary diffusion amounts to estimating the parameters of a stochastic differential equation whose stationary distribution matches a target distribution. We build on the recently introduced kernel deviation from stationarity (KDS), which enforces stationarity by evaluating expectations of the diffusion’s generator in a reproducing kernel Hilbert space. Leveraging the connection between KDS and Stein discrepancies, we introduce the Stein-type KDS (SKDS) as an alternative formulation. We prove that a vanishing SKDS guarantees alignment of the learned diffusion’s […]

Ver mais