✳flâneur — a map of the web's best reading

Tensorflow deep retrieval using Two Towers architecture | Google Cloud Blog

cloud.google.com · 3,808 words · saved by 1 readers

In a previous blog, we outlined three approaches for implementing recommendation systems on Google Cloud, including (1) a fully managed solution with Recommendations AI, (2) matrix factorization from BigQuery ML, and (3) custom deep retrieval techniques using two-tower encoders and Vertex AI Matching Engine. In this blog, we dive deep into option (3) and demonstrate how to build a playlist recommendation system by implementing an end-to-end candidate retrieval workflow from scratch with Vertex AI. Specifically, we will cover: All related code can be found in this GitHub repository. To meet low latency serving requirements, large-scale recommenders are often deployed to production as multi-stage systems. The goal of the first stage (candidate retrieval) is to sift through a large (>100M elements) corpus of candidate items and retrieve a relevant subset (~hundreds) of items for downstream ranking and filtering tasks. To optimize this retrieval task, we consider two core objectives: Two-t

AI & Machine Learning Scaling deep retrieval with TensorFlow Recommenders and Vertex AI Matching Engine April 19, 2023 Jeremy Wortz ML specialist, Google Cloud Jordan Totten Machine Learning Specialist In a previous blog , we outlined three approaches for implementing recommendation systems on Google Cloud, including (1) a fully managed solution with Recommendations AI , (2) matrix factorization from BigQuery ML , and (3) custom deep retrieval techniques using two-tower encoders and Vertex AI Matching Engine . In this blog, we dive deep into option (3) and demonstrate how to build a playlist r

Explore this link on the map →

related reading