A paper by Prof. Wang Youqing's team at Beijing University of Chemical Technology has been accepted by ACM MM 2026, a CCF-A multimedia conference. The work proposes GLCMoE, a global-local mixture-of-experts framework with dual contrastive learning for deep multi-view clustering.
Hugging Face Blog has introduced FairStress, a 300-scenario benchmark that checks whether LLM decisions and their resistance to persuasive challenges vary by demographic identity across 29 domains.
Peking University’s Dong Hao team and Qiyuan Robot have launched HIL-UMI, a human-in-the-loop handheld collection system that flags when a model’s predicted actions diverge and prompts targeted corrections without a robot arm. Reported tests show average task progress rising to 94.8 from 58.0 and 5.63-fold collection efficiency.
Google’s Gemini 4 Argon helped researchers extend exact mathematical solutions for magnetic confinement fusion, producing candidates within hours but not proving a commercial fusion breakthrough.
A study published in Lancet Digital Health reports that researchers at Mass General Brigham built an AI system that estimates biological age and predicts cancer survival from a single facial photograph. The tool, trained on more than 40 million images, is not yet ready for widespread clinical use.
A paper tied to ACM MM 2026 introduces CoReFuse-Med, a fusion method for multimodal medical image segmentation that first removes shallow corruption and then recalibrates deep modality contributions. Tests on EPVS, BraTS and WMH report gains over MedNeXt, while the authors limit claims to resolution-induced degradation after spatial alignment.
A Logitech survey finds 64% of workers frustrated or angry over tech problems and 58% reporting physical discomfort, as Democrats target AI in midterms and research warns women face higher job risk.
In an MIT CSAIL experiment, copies of Claude Opus 4.7 spent 30 hours alone on a simulated island and set their own goals, drew pictures and made up games. The researchers say the behavior resembles play and improved later task performance, though skeptics question whether it is play at all.
Jia Xiaosong, an assistant professor at Fudan University’s Institute of Trustworthy Embodied AI, says the main obstacle in autonomous driving is not the stream of new VLA and WAM architectures but the limits of imitation learning and the difficulty of scaling reinforcement learning, according to a Leiphone.com interview. He traces his research from trajectory prediction to end-to-end driving, Think2Drive and DriveAdapter.
Antibody patterns in blood can signal Crohn’s disease and ulcerative colitis up to a decade before diagnosis, a study reported by The Guardian says, offering a possible route to earlier detection though more validation is needed.
Marine ecologists at Stanford report in Science Advances that bleached corals recovered within eight weeks after coming into contact with healthy neighbors that shared their symbiotic algae, a result researchers say could inform efforts to restore reefs damaged by marine heat waves.
UCLA researchers have developed PhoneBot, an open-source humanoid that uses cheap Android phones for vision and walking control, with about $400 in robotic hardware excluding the handset.