MACHINE LEARNING
PROJECTS
PEOPLE
PUBLICATIONS
PATENTS
Machine Learning
Tianxiao Li is a postdoctoral scientist in the Machine Learning Department at NEC Laboratories America. He received his undergraduate degree in Biological Sciences from Tsinghua University and earned his Ph.D. from Yale University. His expertise spans deep learning, predictive analytics, generative modeling and computational biology, with a focus on building machine learning models that support real-world decision-making in healthcare and pharmaceuticals.
Since joining NEC in 2024, Tianxiao has been a core contributor to the Physics-Informed Machine Learning Project, advancing new methods for compositional generation and reasoning that have potential applications in areas like drug discovery, diagnostics and personalized medicine. He collaborates closely with NEC’s ML team to integrate physics-informed learning, generative modeling, and explainable AI into systems that are both innovative and reliable.
Tianxiao’s work also addresses key challenges in the biomedical domain: diverse modality, high dimensionality, and noisy and insufficient data, by developing more data and parameter-efficient approaches that leverage prior physical and biological knowledge, reduce research costs, shorten development cycles, ensure safety and reliability, and open new opportunities for the industry.
Overview: Since 2019, the parameters of large deep learning models have increased by over 300 times every 18 months. However, the future ML progress cannot continue simply based on using more data or creating larger models, because the growing gap between the model demand and resource supply is not sustainable.
Designing drug molecules is no longer just about generation, but control. NEC Laboratories America introduces MolDiffdAE, a diffusion-based framework that enables precise, multi-objective tuning of 3D molecular properties. By learning a semantic space, researchers can efficiently guide design, accelerating
Controlled generation of 3D molecules is highly valuable in drug design, enabling targeted modifications that preserve core structure and binding-relevant geometry while improving developability-related properties. While equivariant diffusion models achieve state-of-the-art de novo 3D molecule generation,
Designing protein-binding proteins with high affinity is critical in biomedical research and biotechnology. Despite recent advancements targeting specific proteins, the ability to create high-affinity binders for arbitrary protein targets on demand, without extensive rounds of wet-lab testing,remains
We consider the conditional generation of 3D drug-like molecules with explicit control over molecular properties such as drug-like properties (e.g., Quantitative Estimate of Druglikenessor Synthetic Accessibility score) and effectively binding to specific protein sites. To tackle this problem, we propose
We are thrilled to announce the launch of our latest research initiative, the Trustworthy Generative AI Project. This ambitious project is set to revolutionize how we interact with multimodal content by developing cutting-edge generative models capable of compositional generation and reasoning across
In protein biophysics, the separation between the functionally important residues (forming the active site or binding surface) and those that create the overall structure (the fold) is a well-established and fundamental concept. Identifying and modifying those functional sites is critical for protein