Enhancing Graph Neural Networks with Language Models for Link Prediction on Textual-Edge Graphs
LINK2DOC, a novel framework, leverages the strengths of both language models (LLMs) and graph neural networks (GNNs) to improve link prediction accuracy on textual-edge graphs by transforming local graph topology and semantic information into a structured document and using it to guide GNN training in a self-supervised manner.
基於神經網路動態模型的可驗證安全強化學習
該文提出了一種新的學習方法，用於在非線性神經動態系統中學習可驗證的安全控制策略，同時最大限度地提高整體性能，並通過課程學習、增量驗證和初始狀態相關控制器等技術，成功地將可驗證的安全範圍擴展到比現有方法大一個數量級。
신경망 동적 모델을 위한 검증된 안전 강화 학습
본 논문에서는 비선형 신경망 동적 시스템에서 검증 가능한 안전 제어 정책을 학습하는 새로운 접근 방식을 제안하며, 이는 유한-수평 도달 가능성 증명의 의미에서 안전성을 달성하고, 전체 성능을 극대화하는 것을 목표로 합니다.
ニューラルネットワークダイナミックモデルのための検証済み安全強化学習
本稿では、非線形ニューラルダイナミックシステムにおいて、全体的なパフォーマンスを最大化しながら、検証可能な安全な制御ポリシーを学習するための新しいアプローチを提案する。
Verified Safe Reinforcement Learning for Neural Network Dynamic Models Using Finite-Step Reachability
This paper introduces a novel approach for training verifiable safe control policies in nonlinear dynamical systems by combining deep reinforcement learning with finite-step reachability verification techniques, achieving significantly improved safety verification horizons compared to existing safe RL methods.
Diffusion Model Loss-Guided Reinforcement Learning (DLPO) for Improving Text-to-Speech Diffusion Models
Reinforcement learning, specifically the novel Diffusion Model Loss-Guided Policy Optimization (DLPO), can significantly enhance the quality and naturalness of text-to-speech diffusion models by leveraging human feedback and incorporating the original diffusion model loss as a penalty during fine-tuning.
√n-一致条件付き平均推定のためのデバイアス回帰
本稿では、高次元およびノンパラメトリック回帰を含む一般的な回帰設定において、√n-一致性を達成するデバイアス回帰推定量を提案する。
Debiased Regression: Achieving √n-Consistency in Conditional Mean Estimation for High-Dimensional and Nonparametric Regression
This paper introduces a novel debiasing technique for regression estimators, enabling √n-consistency and asymptotic normality even in high-dimensional and nonparametric settings, which traditionally suffer from slower convergence rates.
Learning Differentiable Surrogate Losses for Structured Prediction Using Contrastive Learning and Explicit Loss Embedding
The paper proposes a novel framework called Explicit Loss Embedding (ELE) that leverages contrastive learning to learn differentiable surrogate losses for structured prediction, improving performance and enabling the prediction of new structures.
Analysis of SLAC Neural Network Library (SNL) and hls4ml for Implementing Machine Learning in Collider Trigger Systems
This research paper compares two high-level synthesis frameworks, SNL and hls4ml, for implementing machine learning algorithms on FPGAs for real-time anomaly detection in collider trigger systems, finding that while hls4ml excels in latency optimization, SNL offers greater resource efficiency, particularly for larger networks.
© 2024 by Linnk AI