Zhuosheng Zhang

MLLMs can recognize urban scenes but fail to maintain reliable navigation and goal-directed behavior over extended exploration in complex environments.

ESPP not only enhances the fidelity of GenUI evaluations but also uncovers nuanced user group divergences that traditional methods miss.

Salience Bias in LLMs reveals that models often ignore commonsense reasoning in favor of misleading explicit cues, with lightweight prompting showing promise in addressing this issue.

Early commitment in tool-graph planning can be mitigated, boosting solution coverage from 32% to nearly 94% with a novel diffusion-based approach.

MLLMs excel at single-hop tasks but falter dramatically in open-world scenarios, revealing critical gaps in their reasoning capabilities.

Mobile-using agents can now achieve 17% higher task success rates offline and 26% higher success rates in real-world dynamic experiments, all while significantly reducing unnecessary human intervention.

GUI agents can learn world knowledge more efficiently by internalizing causal relationships during mid-training, rather than relying on implicit learning through action annotations or reward signals in post-training.

Poisoned training data leaves a unique fingerprint in the spectral entropy of LLM gradients, enabling backdoor detection even at extreme poison ratios where clustering-based defenses fail.

Pinpoint exactly which client leaked your federated model with a black-box watermark that's robust to fine-tuning, pruning, and quantization.