Online clothing stores typically recommend products by looking at their customers’ past purchases or searches, and then suggest items that look similar to those products. Perhaps that’s a good strategy when you’ve only been searching for, say, a striped shirt and haven’t bought one yet. But what if you’ve just bought a striped shirt — what’s the likelihood that you’re going to buy another one? What if the store showed you a skirt that matched perfectly with that shirt instead? Wouldn’t that be more helpful?

In Part 1 of this series, we discussed how transfer learning allows you to enjoy the benefits of deep learning while avoiding the high training costs and effort. Using our Custom Collection API, which works based on this concept of transfer learning, we built a fashion matching demo. In this video, created in partnership with DeepLearning.TV, we’ll take a closer look at how e-commerce fashion sites can improve product recommendations to provide a better experience for shoppers, as well as boost likelihood to purchase.

If you want to learn how to build this fashion matching model yourself, check out the tutorial in our final installment of this series. Visit https://indico.io/demos/clothing-matching if you want to play with the demo!

About Custom Collections

Our Custom Collections API allows you to build customized machine learning algorithms with 100x less data than starting from scratch. Check out our docs to learn more.

P.S. We’re also holding a competition this month where you can win 100k API calls (and more) for building an image classifier with Custom Collections!

Suggested Posts

Recognizing Emotion in Text with Machine Learning (No Code Required) 😠😂😨😢😮

Inspiring Hack Projects: ContextCam

indico Wins Boston TechJam 2015