Cho Ho Lam

Title: Cho Ho Lam and His Patent on Neural Network Explanations.

Introduction

Cho Ho Lam is an inventor associated with Burnaby, Canada, and Huawei Cloud Computing Technologies. His listed patent focuses on making the reasoning behind neural network decisions more understandable.

Latest Patents

Lam is credited with 1 patent: “Methods, Systems, and Media for Contextual Discriminative Explanation of Convolutional Neural Networks.” The invention describes methods for explaining binary inferences made by a neural network. It identifies relationships between detected features and the data samples containing them, then pairs features that establish an inference context with features that form the basis for an inference. The system can generate human-perceptible explanations, find similar samples sharing both the context and basis, and identify likely alternatives that share the context but not the basis.

Career Highlights

The available information associates Lam with Huawei Cloud Computing Technologies and identifies his work on explainable neural network behavior as a patent contribution. The invention addresses how people can better understand the basis for a model’s binary inference.

Collaborations

Yong Zhang is listed as Lam’s coworker. The available details do not specify their roles in the patent or describe any particular collaboration.

Conclusion

Cho Ho Lam’s listed patent explores ways to make convolutional neural network inferences more interpretable by presenting the context and features behind a model’s decisions.