Fly0 introduces a structured semantic-geometric interface for zero-shot UAV navigation, achieving over 20% success rate improvement and halving localization error.

Zhenxing Xu, Yihong Lu, Weidong Bao et al.

Proposes Agentic Time Series Forecasting (ATSF) emphasizing perception, planning, action, reflection, and memory.

Mingyue Cheng, Xiaoyu Tao, Qi Liu et al.

Autoregressive image generation method using diffusion loss improves condition errors; experiments show superiority over existing models.

Yucheng Zhou, Hao Li, Jianbing Shen

Mind-Brush enhances image generation with a dynamic knowledge-driven workflow, boosting Qwen-Image performance on Mind-Bench from 0.02 to 0.31.

Jun He, Junyan Ye, Zilong Huang et al.

ObjEmbed achieves fine-grained image-text alignment through multimodal object embeddings, excelling on 18 benchmarks.

Shenghao Fu, Yukun Su, Fengyun Rao et al.

Using GPT-4 to assign weekly gender scores to 70,000+ LLM agents, revealing persistent gender homophily driven by social selection and influence mechanisms.

Faezeh Fadaei, Jenny Carla Moran, Taha Yasseri

PISCES uses dual optimal transport-aligned rewards for annotation-free text-to-video post-training, enhancing video quality and semantic consistency.

Minh-Quan Le, Gaurav Mittal, Cheng Zhao et al.

PENCIL employs a plain Transformer with sampled local subgraphs for link prediction, achieving high accuracy with significantly fewer parameters.

Quang Truong, Yu Song, Donald Loveland et al.

Proposes Rubric-ARM, an alternating RL framework jointly optimizing rubric generator and judge, achieving 4.7% improvement in reward modeling accuracy.

Ran Xu, Tianci Liu, Zihan Dong et al.

Decomposition-based causal discovery (DCD) separates trend, seasonal, residual components, improving causal inference in non-stationary time series by 20-30% over baselines.

Muhammad Hasan Ferdous, Md Osman Gani

A-MapReduce optimizes wide search tasks using a MapReduce-inspired framework, achieving 5.11%-17.50% Item F1 improvements.

Mingju Chen, Guibin Zhang, Heng Chang et al.

ContextEvolve uses three agents to compress optimization context, improving ADRS performance by up to 33.3% while cutting tokens by 29.0%.

Hongyuan Su, Yu Zheng, Yong Li

DeALOG employs a decentralized multi-agent framework with shared natural language logs, achieving robustness and transparency in multimodal QA.

Abhijit Chakraborty, Ashish Raj Shekhar, Shiven Agarwal et al.

Constitutional Evolution uses multi-island LLM evolution; C* reaches S=0.556, +123% over HHH.

Ujwal Kumar, Alice Saito, Hershraj Niranjani et al.

ACE-Step v1.5 uses hybrid architecture for efficient music generation, achieving 100× speedup and running locally on consumer hardware.

Junmin Gong, Yulin Song, Wenxiao Zhao et al.

RecGOAT integrates GNN and OT for dual-level semantic alignment, achieving state-of-the-art multimodal recommendation performance.

Yuecheng Li, Hengwei Ju, Zeyu Song et al.

Proposes BDG framework combining MAS-GLCM and diffusion models for multi-task universal image restoration, significantly improving fidelity.

JiaKui Hu, Zhengjian Yao, Lujia Jin et al.

ConLA employs contrastive disentanglement to learn pure latent actions from human videos, surpassing robot trajectory pretraining with only video data.

Weisheng Dai, Kai Lan, Jianyi Zhou et al.

Proposes Stackelberg game-based reward shaping to optimize LLM alignment, boosting user utility by over 66% win-tie rate.

Haichuan Wang, Tao Lin, Lingkai Kong et al.

PCBSchemaGen uses reward-guided structured verification for training-free PCB schematic generation, achieving 81.3% success rate across 227 real IC tasks.

Huanghaohe Zou, Peng Han, Emad Nazerian et al.