Skip to main content

Postdoc Portrait: Bo Zhao Investigates Neural Networks for Reliable AI

This postdoctoral researcher explores mathematical symmetry in neural networks to make modern artificial intelligence systems more interpretable, efficient, and reliable.

Brought to you byThe Scientist
| 3 min read
Headshot of postdoctoral Bo Zhao.
Register for free to listen to this article
Listen with Speechify
0:00
3:00

Bo Zhao is a postdoctoral researcher at Harvard University. She uncovers mathematical structures in neural networks to understand AI models and improve their reliability. In this Postdoc Portrait interview, she shares the importance of her work and what she hopes her research will accomplish.

Uncovering Neural Network Architecture and Symmetry

Q | What scientific problem are you trying to solve?

I uncover mathematical structures in neural networks to understand AI models.

Modern neural networks can be remarkably capable, but we still lack a clear map of how they work internally. I study one source of this difficulty: the same network function can often be represented by many different settings of its parameters. This matters because two functionally identical models can look very different internally, which can confuse attempts to compare, interpret, or improve them.

I use symmetry, the study of changes that leave something unchanged, to organize these equivalent versions of neural networks. My goal is to build a mathematical language that characterizes this redundancy, then use this structure to design better methods for optimization, model merging, compression, and interpretability. More broadly, I want to make modern AI systems easier to analyze, more efficient to train, and more reliable to use.

Q | What drew you to understanding AI models?

I was drawn to the way mathematics can reveal simple structure beneath complicated behavior. Neural networks are a striking example: they work extremely well, yet even basic questions about why training succeeds or what their internal representations mean remain unsettled.

Continue reading below...

Like this story? Sign up for FREE Newsletter updates:

Latest science news storiesTopic-tailored resources and eventsCustomized newsletter content
Subscribe

Symmetry offered a natural bridge. In mathematics and physics, symmetry often reveals what remains essential when an object is changed in appearance. Neural networks have similar features. Their internal weights can change in systematic ways while their outputs stay the same. I was fascinated that this familiar principle could help explain practical phenomena in machine learning, including why there are many equally good solutions, how trained models can be connected or combined, and why equivalent models can learn differently.

The field lets me move between proofs and experiments, and between foundational questions and practical algorithms. That combination continues to motivate me.

Q | What’s one thing you learned from your research that you didn’t expect?

One unexpected lesson was that redundancy is not always something to eliminate. I initially thought of neural-network symmetries mainly as multiple ways of writing the same model. However, two parameter settings can produce the same outputs while behaving differently during training: their gradients and the local shape of the optimization problem can change.

This means symmetry can be used as a tool. We can move a model to an equivalent internal form that is easier to optimize or use symmetry to test whether an interpretation depends on an arbitrary coordinate choice. More broadly, I learned that “the same model” can mean different things depending on whether we care about its outputs, its training process, or its internal explanation. Making those distinctions precise has opened more questions than I expected.

Q | If your research succeeds, what could it change for science or society?

Scientifically, this work could provide a common mathematical language for understanding neural-network architecture, training, and interpretation. Practically, it could help us choose functionally equivalent versions of a model that train faster, merge more cleanly, compress more accurately, or are easier to interpret. For society, the most important benefit may be more trustworthy evaluation of AI systems. If an interpretability or safety method changes its conclusion only because the same computation is represented differently, it can create false confidence. Mathematical tools that identify stable structure could make such analyses more reproducible and harder to fool. Symmetry alone will not solve AI reliability, but it can provide a rigorous foundation for deciding which internal measurements deserve our trust.

Q | What question are you most excited to answer next?

Today's most capable AI systems are often built by training at scale and observing what emerges. I want to uncover the underlying structures that govern how these systems learn, organize information, and make decisions, then use those structures to design models that are more efficient, interpretable, and reliable. I hope to help move AI from an empirical craft toward a mature science, one in which we can explain not only what a model does, but why it does it and how to make it behave better.


Responses have been edited for length and clarity.

Add The Scientist as a preferred source on Google

Add The Scientist as a preferred Google source to see more of our trusted coverage.

Related Topics

You might also be interested in...
Loading Next Article...
You might also be interested in...
Loading Next Article...
November digest cover
November 2026

Chemistry Meets Arcade Combat in “Reaction Rumble” Game

Choose your fighter! Enter the arena as chemical elements and dole out moves inspired by real-world reactions in this educational browser-based game.

View this Issue
Thumbnail

Make It Local, Make It Last: The Evolution of mRNA Protein Replacement Therapy

Engineering Bispecific Antibodies Against Emerging Viruses

Engineering Bispecific Antibodies Against Emerging Viruses

innovative-research-horizontal-lock-inverted-rgb-900px-w-72ppi
Advancing Research With Humanized Mouse Models

Advancing Research with Next-Generation Humanized Mouse Models

bioxcell
3d illustration showing amyloid plaques in Alzheimer's disease

Best Practices for Blood-Based Amyloid Pathology Testing

fujirebio logo

Products

Abstract image of circles and concentric lines on a blue and pink background.

MX Analytical and Precision Balances: Effortless Weighing and Secure Data Handling

Mettler
3D multicolored conceptual image of antibody molecules

Premium BsAb: LC-MS-Assisted Bispecific Antibody Production with >98% Heterodimer Purity

WuXi Biologics
Abstract conceptual image of AI technology for medical science research.

Exploring a New Era of Pharmaceutical Innovation

A photograph of plastic fragments floating in water, symbolizing contamination from sample storage tubes.

LVL SAFE® 2D-Coded Tubes: The Perfect Tube for Your Sample

AutoGen