Ting Hua

I’m an Assistant Research Professor at the University of Notre Dame. My research is driven by several questions:

• Reasoning and LLM architecture: Can LLMs evolve by themselves? What are the learning dynamics between data and model capabilities? Beyond transformers, are there better architectures?
• Efficient learning: Can small models achieve comparable performance with large models? How can we train models continually without forgetting? Can we achieve effective adaptation with fewer trainable parameters?
• LLM evaluation and explanation: How do we reliably evaluate LLMs, and how can we explain their behavior?
• Interdisciplinary AI applications: Can LLMs accelerate scientific discovery (drug design, computational chemistry)? How can AI enhance teaching and learning? How can AI address social challenges and support vulnerable populations?

Previously, I was a research scientist at Samsung Research America, where I worked on model compression and continual learning with applications in natural language understanding (NLU) and automatic speech recognition (ASR). I received my PhD from Virginia Tech, focusing on probabilistic graphical models and its application on social events detection.