See-Kiong Ng

WorldEvolver redefines LLM agent planning by achieving unprecedented prediction accuracy and decision-making success through self-evolving memory mechanisms.

Noise in multi-behavior recommendation can be effectively mitigated through a novel spectral filtering approach that enhances representation purity and reliability.

FineVerify boosts GPT-5-mini's accuracy by 8.2 points with just four sampled trajectories, outperforming standard scaling methods.

LLM-based recommendation systems can now dynamically adjust the granularity of knowledge graph retrieval, boosting performance by adapting to the complexity of user queries.

Weak-to-strong reward models can ace the test but still fail in the real world, revealing a hidden brittleness in current preference learning approaches.

The fragmented field of world modeling can now be unified under a "levels x laws" taxonomy, revealing critical gaps in autonomous model revision and decision-centric evaluation.

LLMs struggle to maintain context and avoid distraction when reasoning about causality, leading to a significant performance drop as tasks increase in complexity.

Forget hand-coded strategies: METRO uses LLMs to automatically learn dialogue strategies from expert transcripts, achieving state-of-the-art results in non-collaborative dialogue.

Customer service chatbots can be transformed from reactive support tools into proactive business intelligence engines by strategically probing users for information.

A cleverly shaped and rendered adversarial patch can reliably spoof palmprint recognition, even across different models and datasets.

FreeAct boosts quantized LLM performance by dynamically adapting activation transformations to different token types, moving beyond the static transformations that limit existing methods.