Application of semi-supervised neural net for eCommerce image classification

A talk by Binwei Yang
Distinguished Data Scientist, Walmart Global Tech

Register to watch this content

By submitting your email you agree to the Terms of Service and Privacy Statement
Watch this content now

Stages covered by this talk

About this talk

Semi-supervised image classification leverages unlabeled data as well as labelled data to increase classification performance. We built a plug-and-play platform that unifies the pretraining of neural net encoders using unlabeled data and automated search of these Semi-Supervised Learning (SSL) encoders for image classification tasks.

Our ultimate design goals are: • Pretrain models with unlabeled Walmart catalog images using SSL and SoTA neural architectures • With a library of pretrained SSL models, enable a class of downstream classification tasks with better performance

We will discuss practical challenges expanding SSL research such as SimCLR, BYOL and SwAV to work with suitable neural nets beyond ResNet. We will give an overview of the multi-stage image classification framework and share lessons learned from eCommerce use cases.

Have you got yours yet?

Our All-Access Passes are a must if you want to get the most out of this event.

Check them out

Proudly supported by

Want to sponsor this event? contact us.