GPU-Accelerated Genetic Programming: How Does Population Size Affect Performance?
A latest arXiv paper explores how population size affects training performance in GPU-accelerated genetic programming un…
📂 Research
Latest articles in Research
IAM: Identity-Aware Joint Generation of Human Motion and Body Shape
A latest arXiv paper proposes the IAM framework, which for the first time integrates human identity characteristics—body…
XGBoost-Driven Lookup Tables Enable Lightweight Real-Time Rendering Optimization
A latest arXiv paper proposes a lightweight approach based on XGBoost-driven lookup tables, achieving an optimal balance…
New Study Proposes Multi-Domain Retail Receipt OCR Adaptive Enhancement Benchmark Framework
A latest arXiv paper proposes a quality-aware adaptive OCR pipeline that systematically benchmarks receipt digitization …
ResetEdit: Resettable Starting Latent Enables Precise Image Editing
A latest arXiv paper proposes ResetEdit, a method that introduces a "resettable starting latent" mechanism to achieve a …
XTC-Bench: The First Cross-Task Consistency Benchmark Challenging Unified Multimodal Models
Researchers introduce XTC-Bench, the first benchmark to systematically evaluate semantic consistency between visual unde…
ViPO: A New Paradigm for Visual Preference Optimization at Scale
A research team introduces the ViPO framework and Poly-DPO algorithm, tackling the challenges of data noise and conflict…
Semi-Supervised Learning Cracks the Preference Noise Problem
A new study proposes a semi-supervised learning approach to optimize DPO training, theoretically revealing the noise pro…
Neural Cellular Automata: A Completely New Network Architecture?
A latest arXiv survey paper systematically reviews the development of Neural Cellular Automata (NCA), provides reference…
One Perturbation, Two Failure Modes: New Research on VLM Typographic Injection Safety
A latest arXiv paper proposes an embedding-guided typographic perturbation method, systematically revealing two failure …
M³-VQA: A New Benchmark for Multi-Modal, Multi-Entity, Multi-Hop Reasoning Visual Question Answering
A research team has released the M³-VQA benchmark, focusing on multi-entity and multi-hop reasoning visual question answ…
DouC: Dual-Branch CLIP Enables Training-Free Open-Vocabulary Segmentation
Researchers propose the DouC framework, which achieves high-quality open-vocabulary semantic segmentation without additi…