2874 papers published on ArXiv in the cs* category. 311 excluded by clustering as noise.

311 clusters identified with an average of 8.18 papers

Largest clusters:

  1. Advances in Numerical Methods for Complex Systems - 36 papers
  2. Advances in Cybersecurity and Software Vulnerability Detection - 31 papers
  3. Advances in Explainable AI and Interpretable Machine Learning - 28 papers
  4. Advances in Wireless Communication Systems - 27 papers
  5. Advances in Large Language Models for Complex Applications - 22 papers
  6. Advances in Video Generation and Control - 21 papers
  7. Advances in Graph Algorithms and Network Analysis - 20 papers
  8. Advancements in Large Language Models for Medical Diagnosis and Healthcare - 20 papers
  9. Advancements in Assistive Technologies for Visually Impaired Individuals - 20 papers
  10. Advances in Medical Image Segmentation - 20 papers

41 clusters of clusters identified with an average of 59.95 papers

Largest clusters:

  1. Advances in Large Language Models and Reasoning Capabilities - 109 papers
  2. Advances in Point Cloud Processing, Computer Vision, and Related Fields - 105 papers
  3. Advances in Multimodal Learning and Artificial Intelligence - 100 papers
  4. Advances in Numerical Methods and Scientific Computing - 89 papers
  5. Advances in Robotics and Artificial Intelligence - 86 papers
  6. Advances in Computational Complexity, Quantum Computing, and Graph Neural Networks - 83 papers
  7. Advances in Safety-Critical Systems and Cybersecurity - 83 papers
  8. Advances in Human-Centric Artificial Intelligence - 78 papers
  9. Advances in Video Understanding and Processing - 77 papers
  10. Advances in AI-Driven Scene Understanding and Generation - 73 papers

Advances in Large Language Models and Reasoning Capabilities - 109 papers

Researchers have made significant breakthroughs by integrating large language models with external search processes and developing more efficient methods for evaluating their performance. Notable papers like FastCuRL and ConSol have also proposed innovative approaches to accelerate training efficiency and reduce computational costs.

Advances in Point Cloud Processing, Computer Vision, and Related Fields - 105 papers

Researchers are developing novel methods, such as UniPCGC and MagicColor, to enhance point cloud compression and color transfer. Large language models and techniques like Gaussian Splatting are also being leveraged to improve design efficiency, scene understanding, and object manipulation in various applications.

Advances in Multimodal Learning and Artificial Intelligence - 100 papers

Researchers have made significant breakthroughs in developing more efficient and versatile models, such as Mixture-of-Experts models, for tasks like visual tracking and multimodal information acquisition. Notable papers have demonstrated state-of-the-art results in areas like text-to-image generation, multimodal learning, and medical image analysis, showcasing promising applications in human-computer interaction and healthcare.

Advances in Numerical Methods and Scientific Computing - 89 papers

Researchers have developed innovative methods such as Neural Operator-Based Flow Surrogates and hybrid architectures combining Fourier neural operators and convolutional neural networks to improve flow simulations and solve partial differential equations. These advances have the potential to significantly impact fields like fluid dynamics, materials science, and biomedicine with faster, more accurate, and efficient simulations and computations.

Advances in Robotics and Artificial Intelligence - 86 papers

Researchers are developing innovative approaches, such as adaptive interaction mechanisms and latent-to-latent policies, to create more robust and adaptive robots. These advancements enable robots to learn from experience, adapt to new tasks, and interact safely and efficiently with humans in complex social environments.

Advances in Computational Complexity, Quantum Computing, and Graph Neural Networks - 83 papers

Researchers have developed new algorithms, such as a 3.3904-competitive online algorithm and a quantum constraint generation framework, to solve complex problems efficiently. The introduction of certified defenses, like PGNNCert, has also enhanced the robustness of graph neural networks against arbitrary perturbations.

Advances in Safety-Critical Systems and Cybersecurity - 83 papers

Graph-based models and neural networks are enhancing vehicle safety and cybersecurity through improved crashworthiness analysis and intrusion detection. Innovative control methods, such as control barrier functions and probabilistic neuro-symbolic layers, are also ensuring safety and stability in complex systems.

Advances in Human-Centric Artificial Intelligence - 78 papers

Innovative methods such as Dual-Level Control and diffusion-based models have improved accuracy and realism in applications like image editing and human motion synthesis. Advances in Spiking Neural Networks, transformer architectures, and event-based vision have also enhanced efficiency and performance in various human-centric AI applications.

Advances in Video Understanding and Processing - 77 papers

Neural compressors and lattice coding have achieved optimal rate-distortion-perception tradeoffs, while large language models and multi-modal approaches have improved action recognition and video representation learning. Novel methods such as frame selection strategies and self-reflective sampling have enhanced the efficiency and accuracy of long video understanding.

Advances in AI-Driven Scene Understanding and Generation - 73 papers

Researchers have introduced innovative models such as LLM-SAP and DreamLLM-3D, which leverage large language models and vision-language models to improve surgical action planning and 3D generation. Notable papers like ORION and MotionDiff also propose novel methodologies for autonomous driving and 4D content editing, showcasing significant advancements in AI-driven scene understanding and generation.

Advances in AI and Computer Vision for Autonomous Systems and Beyond - 72 papers

Researchers are developing more efficient and accurate models, such as novel neural network architectures and vision-language models, to enhance performance in various domains. These innovations aim to reduce computational costs and improve accuracy in applications like autonomous driving, medical imaging, and remote sensing.

Continual Learning and AI Advancements - 71 papers

Researchers have proposed novel methods like SEKF, CSQN, and KAC to improve continual learning, while innovations in diffusion models and transformers have enhanced video and image generation. New approaches like RoSE and CREATE have also been developed to mitigate catastrophic forgetting in class-incremental learning.

Advances in Safe and Alignable Large Language Models - 67 papers

Researchers are developing methods like sparse encoding and multi-objective optimization to improve model safety and alignment. Innovations in fine-tuning, evaluation benchmarks, and personalized models are also emerging to address issues like bias, fairness, and robustness.

Advances in Machine Learning, Causal Discovery, and Artificial Intelligence - 63 papers

Researchers are leveraging large language models and retrieval-augmented generation to improve causal graph construction, leading to more accurate and interpretable results. The integration of causal reasoning with multi-agent reinforcement learning is also gaining traction, enabling more effective coordination and decision-making among autonomous agents.

Advances in Large Language Models and Their Applications - 62 papers

Researchers have developed methods to improve the explainability and transparency of large language models, enabling them to learn from errors and adapt to dynamic environments. Novel approaches, such as bias detectors and agentic frameworks, have also been proposed to address issues of bias and fairness in AI-driven knowledge retrieval and recommendation systems.

Advances in Document Intelligence, Language Models, and Data Processing - 59 papers

Researchers have developed innovative models and techniques, such as PP-DocLayout and LoRA, to improve document intelligence and language models. These advancements enable more efficient, scalable, and robust solutions for tasks like document layout analysis, language modeling, and data processing.

Advances in Reconfigurable Intelligent Surfaces and Edge Computing - 59 papers

Researchers have proposed frameworks like RAISE and ROUTE to optimize RIS placement, resource allocation, and task offloading in edge computing and vehicular networks. These innovations, combined with machine learning and UAV-assisted systems, are enhancing wireless communication systems' efficiency, reliability, and performance.

Advances in Game Theory, Formal Methods, and Distributed Optimization - 58 papers

Researchers are developing novel models and algorithms to address complex interactions and improve cooperation dynamics in fields like game theory and federated learning. New approaches, such as adjusted control policies and Byzantine-robust frameworks, are enhancing robustness, scalability, and efficiency in distributed systems and optimization.

Advances in Multimodal Large Language Models and Vision-Language Understanding - 58 papers

Researchers are developing modular pipelines, contrastive learning frameworks, and novel distillation techniques to improve visual understanding and perception capabilities in large language models. Innovative approaches, such as generative models and geometry-aware architectures, are being explored to enhance performance in tasks like image retrieval, visual math problem-solving, and assistive technologies.

Advances in Interconnected Research Areas - 57 papers

Transformer-based approaches and semi-supervised learning techniques are improving accuracy and efficiency in fields like sign language recognition and geophysical inversions. Novel methods, such as kinematic information and motion gesture primitives, are also enhancing the realism and accuracy of systems, like those used in time series forecasting and predictive modeling.

Advances in Reinforcement Learning and its Applications - 57 papers

Researchers are proposing innovative methods to mitigate challenges in reinforcement learning, such as reward hacking and sparse rewards. Novel frameworks and algorithms, including parallel computing and GPU acceleration, are improving performance and efficiency in areas like robotic control and complex systems optimization.

Advances in Stochastic Dynamics, Deep Learning, and Complex Systems - 56 papers

Researchers are developing novel frameworks and models to analyze complex phenomena, including new approaches to stochastic dynamics and deep learning. Notable advancements include the introduction of Malliavin-Bismut Score-based Diffusion Models, unified backdoor detection frameworks, and causality-aware methods for predicting human mobility and behavior.

Advances in Domain-Specific Language Modeling and Beyond - 56 papers

Researchers are developing domain-specialized language models, such as OmniScience, and exploring fine-tuning strategies for speech recognition models to improve accuracy in specialized domains. Innovations in audio compression, speech enhancement, and low-resource language processing are also emerging, with a focus on neural network-based approaches and cross-lingual transfer learning.

Advances in Privacy-Preserving Machine Learning and Innovative Computing Solutions - 55 papers

Researchers have developed innovative solutions such as learnable data perturbation and generative adversarial networks to protect sensitive data and prevent malicious attacks. Autonomous optical neural networks and large language models are also being explored to improve efficiency, scalability, and privacy in various computing systems.

Advances in Computer Vision: Towards Robust and Flexible Methods - 54 papers

Model-free frameworks like Any6D and deep learning-based approaches have achieved state-of-the-art results in 6D object pose estimation and keypoint matching. The use of normalized matching transformers and novel loss functions has also improved the robustness and generalizability of pose estimation models, enabling more accurate 3D estimation and disentanglement of camera and scene geometry.

Advances in Large Language Models for Clinical Research and Natural Language Processing - 54 papers

Large language models (LLMs) are improving clinical trial adjudication, data extraction, and text analysis, while also enhancing anomaly detection and natural language processing tasks. Researchers are exploring new LLM applications, including detecting machine-generated text, analyzing temporal relations, and identifying similarities between entities.

Advances in Uncertainty Quantification and Robustness in AI - 53 papers

Researchers are developing new methods to quantify uncertainty in neural networks and creating innovative technologies for human-robot interaction. Large Language Models are being improved to automatically generate fact-checking articles and mitigate hallucinations, enhancing the accuracy and reliability of AI models.

Advances in Human-Centric Technologies - 50 papers

Researchers are developing systems that can analyze user emotions and tailor responses accordingly, creating more immersive and interactive digital humans. Innovations like multimodal transformer models and emotion-sensing systems, such as PERCY and EQ-Negotiator, are enabling more natural and intuitive human-robot interactions.

Advances in AI-Driven Robotics and Mental Health Support - 49 papers

Researchers are using generative AI, language models, and neuro-symbolic learning to improve robot task performance and decision-making. AI-driven tools, such as large language models and interpretable models, are also being developed to generate effective emotional support dialogues and improve mental health support.

Advances in Cybersecurity and AI-Driven Software Development - 48 papers

Large Language Models (LLMs) are being used to enhance security measures, such as autonomous cyberattack detection and vulnerability assessment, and to improve software development through AI-powered code review and bug detection. LLMs are also being applied in programming education to create personalized learning experiences through AI-powered chatbots and immersive virtual reality environments.

Advances in Multimodal Emotion Recognition and Human-Computer Interaction - 47 papers

Researchers are developing frameworks that integrate multiple modalities, such as audio, visual, and physiological signals, to improve emotion recognition accuracy. Deep learning techniques, like convolutional neural networks and contrastive learning, are being used to refine speech emotion recognition and enable more nuanced understanding of human emotions.

Advances in Autonomous Vehicle Research - 43 papers

Cooperative perception approaches have improved perception performance and reduced communication costs in autonomous vehicles. Innovative solutions like federated learning and multimodal sensing frameworks have also enhanced object detection performance and robustness in complex scenarios.

Advances in Memory Management, Distributed Training, and High-Performance Computing - 42 papers

INDIGO achieves 50-70% improvement in application performance with its network-aware page migration framework, while DeFT proposes a communication scheduling scheme that yields speedups of 29% to 115%. Researchers also explore innovative solutions like CIM-aware compression and CIM accelerators, and alternative hardware platforms to reduce energy consumption and increase throughput.

Advances in Decentralized Systems and Blockchain Security - 40 papers

New frameworks and techniques have been developed to assess smart contract reputability and mitigate risks in blockchain systems. Novel architectures and algorithms have also been proposed to improve decentralized identity, access control, and consensus mechanisms, enhancing security, transparency, and governance.

Advances in Artificial Intelligence: Safety, Collaboration, and Ethics - 40 papers

Researchers have developed new testing frameworks like SuperARC to evaluate AI models' ability to generalize and adapt to new situations. AI is being applied to various fields, such as healthcare, and researchers are exploring human-AI collaboration, safety, and regulatory frameworks to mitigate risks.

Advances in Biological Sequence Generation, Protein Design, and Secure Computing - 37 papers

Researchers have introduced novel frameworks for controllable sequence generation and protein design, leveraging integrated multiple modalities. Additionally, innovative methods have been proposed to enhance secure computing, including compression techniques, quantization methods, and homomorphic encryption.

Advances in Audio, Image, and Synthetic Data Security - 37 papers

Researchers have developed innovative methods to detect deepfakes, adversarial attacks, and synthetic media, using techniques such as anomaly detection frameworks and forensic microstructure modeling. Notable papers have introduced novel frameworks, including CAARMA, SITA, CO-SPY, and FakeReasoning, to improve detection and attribution capabilities in audio, image, and synthetic data security.

Advances in Robustness and Generalizability of Deep Learning Models - 35 papers

Researchers have introduced innovative methods, such as the QUBA score and robustness enhancement modules, to improve model robustness and fairness. New approaches, including the Exponentially Weighted Instance-Aware Repeat Factor Sampling method and SILVA framework, have also been developed to enhance model performance and transparency.

Advances in Image Restoration and Generation - 34 papers

Diffusion models have shown significant potential in image restoration and denoising, with applications in blind face restoration and image denoising. Notable developments include novel architectures, frameworks, and techniques, such as latent space super-resolution and contrastive learning, which improve image generation, restoration, and anomaly detection.

Advances in Renewable Energy and Neural Representations - 33 papers

Researchers are developing innovative methods, such as robust control strategies and physics-informed neural networks, to enhance stability and efficiency in renewable energy systems and power grids. Scalable adjoint backpropagation methods and hybrid models are also being created to improve efficiency and accuracy in solving differential equations and neural representations.

Advances in Human-Computer Interaction and AI-Driven Design - 29 papers

Researchers have developed innovative platforms and models for human-AI interaction, media generation, and design interpretation, enabling immersive interfaces and personalized design. These advancements have the potential to revolutionize industries such as education, design, and healthcare with AI-powered frameworks and AI-driven design approaches.

Subsections

Advances in Large Language Models and Reasoning Capabilities

(109 papers)

Advances in Point Cloud Processing, Computer Vision, and Related Fields

(105 papers)

Advances in Multimodal Learning and Artificial Intelligence

(100 papers)

Advances in Numerical Methods and Scientific Computing

(89 papers)

Advances in Robotics and Artificial Intelligence

(86 papers)

Unclustered

(85 papers)

Advances in Computational Complexity, Quantum Computing, and Graph Neural Networks

(83 papers)

Advances in Safety-Critical Systems and Cybersecurity

(83 papers)

Advances in Human-Centric Artificial Intelligence

(78 papers)

Advances in Video Understanding and Processing

(77 papers)

Advances in AI-Driven Scene Understanding and Generation

(73 papers)

Advances in AI and Computer Vision for Autonomous Systems and Beyond

(72 papers)

Continual Learning and AI Advancements

(71 papers)

Advances in Safe and Alignable Large Language Models

(67 papers)

Advances in Machine Learning, Causal Discovery, and Artificial Intelligence

(63 papers)

Advances in Large Language Models and Their Applications

(62 papers)

Advances in Document Intelligence, Language Models, and Data Processing

(59 papers)

Advances in Reconfigurable Intelligent Surfaces and Edge Computing

(59 papers)

Advances in Game Theory, Formal Methods, and Distributed Optimization

(58 papers)

Advances in Multimodal Large Language Models and Vision-Language Understanding

(58 papers)

Advances in Interconnected Research Areas

(57 papers)

Advances in Reinforcement Learning and its Applications

(57 papers)

Advances in Stochastic Dynamics, Deep Learning, and Complex Systems

(56 papers)

Advances in Domain-Specific Language Modeling and Beyond

(56 papers)

Advances in Privacy-Preserving Machine Learning and Innovative Computing Solutions

(55 papers)

Advances in Computer Vision: Towards Robust and Flexible Methods

(54 papers)

Advances in Large Language Models for Clinical Research and Natural Language Processing

(54 papers)

Advances in Uncertainty Quantification and Robustness in AI

(53 papers)

Advances in Human-Centric Technologies

(50 papers)

Advances in AI-Driven Robotics and Mental Health Support

(49 papers)

Advances in Cybersecurity and AI-Driven Software Development

(48 papers)

Advances in Multimodal Emotion Recognition and Human-Computer Interaction

(47 papers)

Advances in Autonomous Vehicle Research

(43 papers)

Advances in Memory Management, Distributed Training, and High-Performance Computing

(42 papers)

Advances in Decentralized Systems and Blockchain Security

(40 papers)

Advances in Artificial Intelligence: Safety, Collaboration, and Ethics

(40 papers)

Advances in Biological Sequence Generation, Protein Design, and Secure Computing

(37 papers)

Advances in Audio, Image, and Synthetic Data Security

(37 papers)

Advances in Robustness and Generalizability of Deep Learning Models

(35 papers)

Advances in Image Restoration and Generation

(34 papers)

Advances in Renewable Energy and Neural Representations

(33 papers)

Advances in Human-Computer Interaction and AI-Driven Design

(29 papers)

Built with on top of