cs.LG
PulseAugur coverage of cs.LG — every cluster mentioning cs.LG across labs, papers, and developer communities, ranked by signal.
- instance of CatalyzeX Code Finder for Papers 70%
- instance of Influence Flower 70%
- instance of IArxiv Recommender 70%
- instance of arXivLabs 70%
- used by cs.RO 70%
- instance of cs.AI 70%
- instance of DL-proline 70%
- instance of cs.RO 70%
- instance of math.OC 60%
- affiliated with Influence Flower 50%
- instance of IArxiv 50%
- affiliated with IArxiv 50%
18 day(s) with sentiment data
-
New bandit framework optimizes stochastic matching with rare updates
Researchers have developed a new bandit framework designed to optimize stochastic matching problems, particularly those involving the multinomial logit (MNL) choice model. This framework addresses the computational chal…
-
New framework SEM enables derivative-free learning for complex systems
Researchers have introduced Sparse Equation Matching (SEM), a new framework designed to discover equations governing complex systems without requiring derivative estimation. This derivative-free approach is applicable t…
-
Research: Compute gap determines adversarial attack transferability
A new research paper explores the relationship between computational resources and the transferability of adversarial attacks in machine learning. The study suggests that a significant computational gap, specifically a …
-
New method uses parallel gradient boosting for nonparametric distribution estimation
Researchers have developed a novel nonparametric autoregressive approach for estimating probability distributions, particularly effective for high-dimensional data. This method leverages conditional distribution estimat…
-
New SABLE method tackles high-dimensional latent space optimization
Researchers have developed SABLE, a novel Bayesian optimization method designed for high-dimensional latent spaces with limited evaluation budgets. This method addresses the challenge where the number of dimensions sign…
-
New research explores causal inference with learned multimodal representations
A new research paper published on arXiv introduces a novel approach to causal inference using learned multimodal representations. The study addresses limitations of current methods that focus on data reconstruction rath…
-
New SAGE audio encoder balances quality, speed, and semantics
Researchers have introduced SAGE, a new Semantic Audio Generative Encoder designed to improve the balance between audio reconstruction quality, semantic structure, and inference speed. This 105M-parameter model, trained…
-
New algorithms tackle resource allocation with censored feedback
Researchers have developed new algorithms, RA-UCB and MG-UCB, to address an online resource allocation problem with censored feedback. This problem involves allocating a budget across multiple arms, where rewards are on…
-
New algorithm offers quasipolynomial time solution for $\ell_\infty$ contractions
A new algorithm has been developed to compute an approximate fixed point for nonexpansions in $\ell_\infty$ norm. This algorithm achieves a runtime of $\log(1/\eps)^{O(\log^2 d)}$, which has implications for solving sim…
-
New ASTEROID framework enhances AI agent exploration efficiency
Researchers have developed ASTEROID, a novel student-teacher framework designed to improve in-context exploration for sequential decision-making agents operating under partial observability. This method trains a student…
-
Silent alignment phenomenon observed in two-layer linear networks
Researchers have analyzed the dynamics of gradient flow within a two-layer linear neural network operating in a high-dimensional setting. Their work, which combines closed-form solutions with random matrix theory, provi…
-
Survey maps reinforcement learning implementations on FPGAs and ICs
This survey paper reviews over 100 studies on the implementation of reinforcement learning (RL) algorithms on field-programmable gate arrays (FPGAs) and integrated circuits (ICs). It categorizes research by RL algorithm…
-
RiboBridge framework designs RNA binders for protein structures
Researchers have developed RiboBridge, a novel framework for designing RNA molecules that can bind to specific protein structures. This structure-first approach separates the generation of RNA conformations from the ass…
-
Looped Transformers: Full Backpropagation Outperforms Truncation in Scaling Study
A new study published on arXiv explores the effectiveness of truncated backpropagation through time (BPTT) in training looped transformers. The research, conducted by Xuyuan Xiong, indicates that while truncation offers…
-
New algorithms for learning positive ReLU networks under Gaussian inputs
Researchers have developed polynomial-time algorithms for learning positive ReLU networks, specifically addressing scenarios with standard Gaussian inputs. The algorithms achieve polynomial sample and real-arithmetic co…
-
Clinical prediction model evaluation methods compared in new paper
This paper compares two methods for evaluating clinical prediction models: decision curve analysis and cost curves. The authors demonstrate a close relationship between decision curves and Brier curves, a specific type …
-
Physics-inspired neural networks tackle graph coloring challenges
Researchers have developed a novel neural network framework that leverages principles from statistical mechanics to efficiently solve complex graph coloring problems. This physics-inspired approach integrates a supervis…
-
Two ReLU Layers Sufficient for Any Neural Network, arXiv Paper Proves
A new paper published on arXiv presents a proof that any Rectified Linear Unit (ReLU) neural network can be precisely simulated by a ReLU network with only two hidden layers. The paper offers two distinct proofs for thi…
-
New bounds established for greedy algorithm in weakly submodular maximization
Researchers Qixin Zhang and others have analyzed the performance of the standard greedy algorithm when applied to weakly submodular maximization problems with matroid constraints. Their work establishes refined lower an…
-
New arXiv papers explore learning theory and physics calibration for AI
Two new research papers have been published on arXiv, focusing on distinct areas of machine learning. The first paper, "Quadratic Sample Complexity for Strictly Replicable Realizable Learning with Unknown Marginals," de…