Xianghao Kong

Hello/你好👋

prof_pic.jpeg

I am a Senior Machine Learning Research Engineer at Apple , where I work on GenAI + Design. I earned my PhD in Computer Science from UC Riverside under the supervision of Prof. Greg Ver Steeg. Fortunately, I also worked at SonyAI(Host: Vikash Sehwag) and Adobe Firefly (Host: Hareesh Ravi) as a Research Intern.

My research centers on Generative Models (Diffusion Models & Energy-Based Models), with a focus on their interpretability, alignment, and compositionality. Specifically, I explore diffusion models through a novel information-theoretic lens, termed Information-Theoretic Diffusion (ITD) ℹ️. Our work demonstrates how Mutual Information enhances compositional reasoning and modality alignment (e.g., text and image).

Before joining UC Riverside, I worked on EEG data analysis for Brain–Computer Interface (BCI) 🧠 technologies, bridging neuroscience and computer science. Outside of research, I enjoy exploring food, sketching, and visiting museums.

news

Jul 13, 2026 A new chapter begins: I’ve joined Apple  to work at the intersection of generative AI and creativity — building technologies that empower artists and designers. I’m also excited to continue exploring research collaborations across industry and academia.
Jun 04, 2026 I wrote a new blog-style note on diffusion models from an information-theoretic perspective, discussing denoising, likelihood estimation, energy, and why density estimation may matter for future interactive generative models.
Nov 17, 2025 I reunited with Michael to create a 2-minute-30-second horror sci-fi short film, Dreamcatcher Hotel. Watch it now and leave likes or comments!
Sep 22, 2025 I’m excited to share that I’ve started a new role as a Video GenAI Researcher at BayArea, where work with Emmy Winners to explore GenAI-powered filmmaking workflows that incorporate real actors. Looking forward to this journey and the creative possibilities ahead!
Sep 03, 2025 PhD defense complete!! Doctor status unlocked 🍻
Apr 26, 2025 I delivered a 20-minute presentation at SOCAMS ☕, had the pleasure of attending many insightful talks, and came away convinced that the REAL AGI still has a way to go.

selected publications

  1. Stretching Each Dollar: Diffusion Training from Scratch on a Micro-Budget
    Vikash Sehwag, Xianghao Kong , and 3 more authors
    2025
  2. Interpretable Diffusion via Information Decomposition
    Xianghao Kong*, Ollie Liu* , and 3 more authors
    2024
  3. Information-Theoretic Diffusion
    Xianghao Kong, Rob Brekelmans , and 1 more author
    2023
  4. ACL
    Asymmetric Bias in Text-to-Image Generation with Adversarial Attacks
    Haz Sameen Shahgir, Xianghao Kong , and 2 more authors
    2024