Page 113 of 130 • 1560 Total Articles

createLiveAI

Continue exploring the latest AI breakthroughs, technology insights, and industry analysis. Page 113 of our comprehensive AI news collection.

All Articles 1560 Business 249 Ethics 150 General 142 Policy 12 Research 793 Startups 13 Technology 201

📰 Latest Intelligence

Showing 12 articles on page 113 of 130

Live feed

📱 2-column layout

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Less is More: Unlocking Specialization of Time Series Foundation Models via Structured Pruning

Research on Time Series Foundation Models (TSFMs) shows that while they achieve strong zero-shot forecasting, fine-tuning often doesn't surpass smaller, specialized models. A proposed \"prune-then-finetune\" approach, which involves structured pruning before fine-tuning, significantly enhances TSFM performance, enabling state-of-the-art results across multiple benchmarks.

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

LLM Agents for Bargaining with Utility-based Feedback

Researchers have developed BargainArena, a new benchmark dataset with six complex bargaining scenarios, to better evaluate large language models' strategic and opponent-aware reasoning. They also introduce a utility-based feedback framework and evaluation metrics inspired by utility theory, which significantly improve LLMs' negotiation performance and alignment with human preferences.

arXiv Machine Learning

Technology

📄 arXiv Machine Learning

May 31, 2025

>-

A new framework, LLM-ODDR, leverages Large Language Models to improve order dispatching and driver repositioning in ride-hailing services by balancing platform revenue, driver fairness, and demand patterns. Experiments with real-world data from Manhattan demonstrate that LLM-ODDR outperforms traditional methods in effectiveness, adaptability, and interpretability, marking a pioneering use of LLMs in transportation decision-making.

technology llm +1

arXiv Machine Learning

Technology

📄 arXiv Machine Learning

May 31, 2025

>-

A novel Decentralized Federated Learning (DFL) framework is proposed to enhance crop disease detection by improving model accuracy, convergence speed, and robustness while preserving data privacy across heterogeneous environments. This approach leverages validation loss for peer model sharing and local training correction, demonstrating superior performance over traditional centralized FL in agricultural applications.

technology machine-learning

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

MAP: Revisiting Weight Decomposition for Low-Rank Adaptation

The paper introduces MAP, a new framework for parameter-efficient fine-tuning of large language models that rigorously decomposes weight adaptation into direction and magnitude by representing weight matrices as high-dimensional vectors, enabling more interpretable and flexible updates. Extensive experiments demonstrate that MAP enhances existing PEFT methods like LoRA, offering a simple, universal approach that can serve as a default for future fine-tuning strategies.

NLP

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Machine Learning Models Have a Supply Chain Problem

The paper highlights the supply-chain risks associated with open machine learning models, such as malicious replacements or training on compromised data, which have already been exploited in attacks. It proposes using Sigstore to enhance transparency by enabling model publishers to sign their models and verify dataset properties, thereby improving security in the open ML ecosystem.

Machine Learning

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Measuring Participant Contributions in Decentralized Federated Learning

This paper introduces new methods for measuring participant contributions in decentralized federated learning (DFL), where clients exchange models directly without a central server. The authors propose DFL-Shapley, an extension of the Shapley value for DFL, and its approximation DFL-MR, both validated through experiments to effectively assess contributions in decentralized settings.

Transformers

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

MermaidFlow: Redefining Agentic Workflow Generation via Safety-Constrained Evolutionary Programming

MermaidFlow is a framework that enhances autonomous agent reasoning by representing workflows as verifiable, human-interpretable graphs and applying safety-constrained evolutionary operators to generate robust, executable plans. This approach improves success rates and convergence speed in workflow generation, offering a scalable and interpretable foundation for reliable agent systems.

Autonomous Systems

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Model-Preserving Adaptive Rounding

Researchers introduce YAQA, an adaptive quantization algorithm for large language models that uses Kronecker-factored Hessian approximations to better preserve the original model's output distribution. YAQA achieves approximately 30% reduction in KL divergence and state-of-the-art performance across various models and tasks.

research machine-learning

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

MoRE: A Mixture of Low-Rank Experts for Adaptive Multi-Task Learning

A new method called Mixture of Low-Rank Experts (MoRE) is proposed to enhance multi-task parameter-efficient fine-tuning of large language models by aligning different LoRA ranks with specific tasks and using an adaptive rank selector, leading to improved performance without extra inference costs. Extensive experiments demonstrate that MoRE outperforms traditional LoRA methods across multiple benchmarks, facilitating more efficient multi-task adaptation of LLMs.

Transformers

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Multivariate de Bruijn Graphs: A Symbolic Graph Framework for Time Series Forecasting

The study introduces DRAGON, an encoder that uses Multivariate de Bruijn Graphs to discretize and represent continuous time series data structurally, facilitating better neural modeling. By integrating graph-based attention into a dual-branch architecture, DRAGON enhances traditional CNN encoders with symbolic, structure-aware features to improve forecasting accuracy.

Transformers

arXiv Machine Learning

Research

📄 arXiv Machine Learning

May 31, 2025

Mustafar: Promoting Unstructured Sparsity for KV Cache Pruning in LLM Inference

The study shows that unstructured sparsity can greatly enhance KV cache compression in large language models, achieving up to 70% sparsity without accuracy loss or fine-tuning. By employing a bitmap-based sparse format and a custom attention kernel, the approach reduces cache size by up to 45%, enabling longer contexts and up to 2.23x faster decoding.

Transformers

1 2 3 4 5 6 7 ... 130

Page 113 of 130 • Showing articles 1345-1356 of 1560

Quick Navigation

Jump to any page or browse by category

Latest (Page 1) Business 249 Ethics 150 General 142 Policy 12 Research 793 Startups 13