Zhaoze Wang

I am a Ph.D. student at the UPenn GRASP Lab. I study predictive world models and compact latent representations for multimodal sequence learning, long-horizon prediction, and planning, drawing inspiration from biological memory and navigation.

I previously interned at Adobe Research, where I worked on video world models, with a focus on efficient generation and long-horizon prediction.

I am seeking research internships from Summer through Fall 2027, with availability for a continuous internship across both seasons.

Email / Google Scholar / GitHub / LinkedIn / X

Research Interests

The brain's ability to build internal predictive models is a key inspiration behind world models. My research asks how such representations can support not just prediction, but fast updating and replanning as observations and goals change.

I study compact predictive representations, recurrent memory, and learned dynamics, drawing on how the brain integrates sensory experience to guide action. My goal is to translate these principles into fast, adaptive robotic systems.

I study how sensory sequences shape spatial representations [1] and how structured predictive states emerge [2]. I also work on memory-guided planning [3] and efficient video world models. To support model training, I develop compact visual encoders [4] and parallel sensory simulation [5].

Selected Projects
	TinyWAM: A Tiny World Action Model Ongoing project · 2026 A compact-token world-action model for goal-conditioned robot motion, with experiments in simulation and optimized inference on a single GPU. Project
	RatatouGym: GPU-Parallel Multimodal Navigation Simulation 2023–Present GPU-parallel simulation combines multi-agent navigation and neural sensor models through Gym-style APIs. Project Page Documentation Source Code

Publications
	A Simple Model of Co-Emergence of Grid and Place Fields Zhaoze Wang, Genela Morris, Dori Derdikman, Pratik Chaudhari^†, Vijay Balasubramanian^† NeurIPS 2026 Self-supervised sensory prediction learns spatial representations from experience and motion. Paper Project Page GitHub
	REMI: Reconstructing Episodic Memory During Internally Driven Path Planning Zhaoze Wang, Genela Morris^†, Dori Derdikman^†, Pratik Chaudhari^†, Vijay Balasubramanian^† NeurIPS 2025 A recurrent model combines cue-triggered goal retrieval, route planning, and sensory reconstruction. Paper Project Page Simulation Suite Vision Encoder
	Time Makes Space: Emergence of Place Fields in Networks Encoding Temporally Continuous Sensory Experiences Zhaoze Wang^†, Ronald W. Di Tullio^†, Spencer Rooke, Vijay Balasubramanian NeurIPS 2024 Reconstructing noisy or partially observed sensory sequences learns spatial representations. Paper Code Video Project Page
	Trading Place for Space: Increasing Location Resolution Reduces Contextual Capacity in Hippocampal Codes Spencer Rooke, Zhaoze Wang, Ronald W. Di Tullio, Vijay Balasubramanian NeurIPS 2024 Oral Geometric analysis quantifies the trade-off between spatial precision and memory capacity. Paper Video

Projects
	BtnkMAE: Compact and Decodable Visual Representations 2023–Present A bottleneck masked autoencoder compresses images into compact embeddings that can be decoded back into images. Source Code
	NN4N: Neural Networks for Neurosimulations 2023–Present Modular recurrent networks with biologically inspired connectivity support flexible, multi-region neural simulations. Documentation Source Code

Teaching & Reviewer

Service

NeurIPS 2025 Reviewer

NeurIPS 2026 Reviewer (Top Reviewer)

Teaching Assistant, ESE 5460: Principles of Deep Learning, Fall 2025

Teaching Assistant, PHYS 5585: Comp. and Theoretical Neurosci., Spring 2026

Design and code of this website is adapted from Jon Barron's website.

Research Interests

Selected Projects

Publications

Projects

Teaching & Reviewer