Combines Quality Diversity (QD) algorithms with supervised discriminative models, using multi-frequency CPPNs and MAP-Elites to explore diverse audio solutions with high novelty and quality.

Björn Þór Jónsson, Çağrı Erdem, Stefano Fasciani et al.

Intervention-aware variational quantum predictive control (IA-VQC-DPC) significantly reduces safety violations and reliance on safety layers in building control, validated via safety attribution protocols.

Yifan Wang

DARP introduces difference-aware retrieval policies, leveraging local neighborhood structures to improve imitation learning robustness, achieving 15-46% performance gains over standard behavior cloning.

Quinn Pfeifer, Ethan Pronovost, Paarth Shah et al.

This paper reveals that RLHF achieves shallow alignment by compressing partisan signals without removing the underlying partisan structure, as shown through internal representation analysis of Llama 3.1 8B.

Wendy K. Tam

AdvGRPO framework combines dense multi-channel rewards and advantage decoupling for joint attacker-defender training, achieving over 90% attack success rate and superior defense robustness.

Blake Bullwinkel, Eugenia Kim, Amanda Minnich et al.

Popcorn benchmark combines title-aligned full-movie/trailer embeddings with VLM-encoded thumbnails to evaluate visual evidence in multimodal movie recommendation.

Ali Tourani, Fatemeh Nazary, Yashar Deldjoo et al.

Agentopia introduces a long-term multi-agent society simulation over 10 years, leveraging life reward-based reinforcement learning to enhance social behaviors and anthropomorphic capabilities of LLMs.

Xintao Wang, Sirui Zheng, Hongqiu Wu et al.

Proposes EmbedFilter, a linear transformation that filters the latent subspace encoding high-frequency, uninformative tokens, improving zero-shot text embedding performance by up to 14%.

Songhao Wu, Zhongxin Chen, Yuxuan Liu et al.

Using a task-based framework, real-world data from Perplexity shows AI agents significantly boost automation, efficiency, and task scope, with productivity gains of up to 87%.

Jeremy Yang, Kate Zyskowski, Noah Yonack et al.

Proposes CoMetaPNS, integrating continual Bayesian GMM with set-conditioned generative models for personalized cardiac electrophysiology simulation, achieving superior accuracy and anti-forgetting.

Ryan Missel, Xiajun Jiang, Linwei Wang

Proposes a physiologically constrained musculoskeletal neural network (MSK-NN) for multi-DoF joint kinematics estimation from partial sEMG, outperforming baseline models.

Wending Heng, Mingming Zhang, Glen Cooper et al.

This study introduces representation steering via Sparse AutoEncoders (SAE) and activation space manipulation to reduce Whisper's hallucination rate from 72.63% to 14.11%, without fine-tuning.

Georgii Aparin, Vadim Popov, Tasnima Sadekova et al.

TEVI leverages sparse autoencoders with text conditioning to refine image embeddings, significantly improving vision-language alignment and retrieval accuracy.

Sweta Mahajan, Sukrut Rao, Jiahao Xie et al.

Proposes Transferability and Predictability as extensions to ISO 26262, enhancing autonomous vehicle controllability and predictability with measurable metrics.

Chaitanya Shinde, Hadi Hajieghrary, Paul Schmitt et al.

This paper introduces a unified framework based on watching, remembering, and reasoning, significantly advancing long video understanding with multimodal LLMs.

Jiahao Meng, Yue Tan, Qi Xu et al.

Proposes a simulation-based imitation learning framework that automatically generates diverse reach-to-grasp demonstrations, achieving over 90% grasp success in real-world tests.

Kaijie Shi, Wanglong Lu, Huiling Chen et al.

This paper introduces LLM-guided MAP-Elites evolution for optimizing medical decision pipelines, improving accuracy and safety metrics significantly across tasks.

Ivan Sviridov, Artem Oskin, Ivan Panin et al.

Proposes a Constrained Dominant Set (CDS) method for multimodal long-document QA, achieving 66.99 on VisDoMBench, surpassing previous SOTA by 37.1 points.

Ambuj Mehrish, Sebatiano Vascon

SES framework optimizes symbolic solutions directly from equations without training data, successfully recovering algebraic and differential equations' explicit expressions.

Sergei Garmaev, Vinay Sharma, Olga Fink

Proposes Code2LoRA, a hypernetwork-based method generating repository-specific adapters, achieving 63.8% cross-repo and 66.2% in-repo exact match on static tasks, and 60.3% on evolving codebases.

Liliana Hotsko, Yinxi Li, Yuntian Deng et al.