Answer Engineering uses local trajectory editing to improve protocol compliance in LLMs, boosting SSNHL compliance from 25.1% to 83.5%.

Victor Lavrenko, Anastasiia Molodnitskaia

Probing LLM preferences via parametric variation reveals significant incoherence.

Elena Ajayi, Angelica Chowdhury, Seth Lazar

Hierarchical Transformer detects multi-turn jailbreaks with 0.9394 F1, outperforming baselines.

Chenhui Hu, Muhammed Salih, Sudipto Guha et al.

Proposes a block-level heterogeneous CPU-GPU inference framework with cross-frame asynchronous pipeline, reducing latency by over 20% on VLA models.

Haibo HU, Lianming Huang, Qiao Li et al.

DUET leverages human demonstration pretraining with Action Chunking Transformer to enhance heterogeneous dual-robot collaboration, achieving 20% higher success rates.

Yiqi Zhao, Ruohai Ge, Celina Shiyu Wang et al.

Proposed ViGiL3D++ combines scene graph sampling and LLM rephrasing to generate diverse 3D grounding descriptions, improving dataset richness.

Austin T. Wang, Dongchen Yang, Angel X. Chang

LPM uses dynamic soft prompts for personalization, improving accuracy by 8.8% on PersonaMem v1, reducing KV-cache by 64x.

Debrup Das, Avinash Amballa, Yashas Malur Saidutta et al.

Vesta unifies localization, navigation, reasoning, and planning in a single model, outperforming state-of-the-art baselines by over 20%.

Johan Bjorck, Zhiqi Li, Yunze Man et al.

JanusMesh is a fast, zero-shot framework for generating dual-semantic 3D illusions using cross-space denoising, completing in 3-5 minutes with high geometric and semantic fidelity.

Siang-Ling Zhang, Huai-Hsun Cheng, Tsung-Ju Yang et al.

UNIEGO employs proxy-mediated hierarchical distillation from nine heterogeneous teachers to unify egocentric video representations, achieving state-of-the-art results.

Wenhao Chi, Arkaprava Sinha, Dominick Reilly et al.

FaraGen1.5 enables scalable data generation for computer use agents using modular components, achieving 72.3% on Online-Mind2Web with Fara1.5-27B.

Ahmed Awadallah, Sahil Gupta, Yash Lara et al.

Proposes Spatially Speculative Decoding (SSD), leveraging 2D spatial prediction to accelerate autoregressive image generation by up to 13.3×.

Shilong Xiang, Zirui Zhang, Lijun Yu et al.

CalTennis is a large multi-view tennis video dataset with over 11 million frames, used to evaluate monocular-to-3D pose estimation, highlighting challenges in depth and foot contact accuracy.

Ilona Demler, Xinran Xie, Blake Werner et al.

Proposes graph-bound execution-state capsules for low-latency, small-batch on-device AI, enabling byte-exact snapshot and restore with sub-millisecond GPU performance.

Liang Su

DeepSWIP leverages neural materialization and quotient WMC for exact single-world counterfactual inference in neural probabilistic logic programs, achieving a 2.14× speedup.

Saimun Habib, Vaishak Belle, Fengxiang He

Using egocentric human videos with a Mixture-of-Transformers model outperforms robot trajectories in embodied pretraining, especially on out-of-distribution tasks.

Juncheng Ma, Jianxin Bi, Yufan Deng et al.

Propose probe-and-refine tuning, an iterative method using synthetic bug probes to enhance repository guidance, improving mean resolve rate from 28.3% to 33.0% with Qwen3.5-35B.

Asa Shepard, Jeannie Albrecht

Integrating Genetic Algorithm and A* with AHP for multi-criteria path planning enhances continuum robot resilience by increasing path diversity and robustness.

Oxana Shamilyan, Ievgen Kabin, Zoya Dyka et al.

Proposes Marginal Advantage Accumulation (MAA) to address cross-batch evidence aggregation, boosting offline knowledge distillation efficiency.

Mingyu Yang, Keye Zheng, Congchao Cheng et al.

Proposes SSH-Net, a deep neural network integrated with cause-specific competing risks model, for failure time distribution prediction on GPU data, leveraging hierarchical data structures.

Jie Min, Yueyao Wang, Mengkun Chen