Zhiting Hu

About

Zhiting Hu is a leading researcher in machine learning, with key contributions spanning deep kernel learning, natural language processing, and generative AI. His work bridges the gap between probabilistic models and deep neural networks, enabling more expressive and scalable solutions for complex data. Hu is perhaps best known for pioneering deep kernel methods that capture recurrent structures in sequential data, a breakthrough with applications in speech, robotics, finance, and biology. His highly cited paper, "Learning Scalable Deep Kernels with Recurrent Structure" (2017, 42 citations), introduced expressive closed-form kernel functions that model ordering and temporal dependencies—a challenge that standard kernels could not address. This work, alongside its earlier version (2016, 22 citations), has shaped how researchers approach sequence modeling in probabilistic frameworks. Hu’s impact extends beyond kernel methods; he has also made notable advances in controllable text generation, reinforcement learning for language models, and interpretable AI. His research is widely recognized for its practical relevance and theoretical depth, earning him a reputation as a versatile and influential voice in the machine learning community. For students and researchers, Hu’s work exemplifies how rigorous mathematical insight can drive real-world innovation.

Research Focus

Key Achievements

Top Papers

1. 1
   Learning Scalable Deep Kernels with Recurrent\nStructure

   42 citations · 2017
2. 2
   Learning Scalable Deep Kernels with Recurrent Structure

   22 citations · 2016

Key Collaborators

Contact & Links