SQL and TensorFlow: Enabling smart queries on a data warehouse
summary
BigQuery plus TensorFlow enables similarity search by converting records into numeric vectors and comparing distances at SQL scale. Examples: TF‑IDF turns 10 million Stack Overflow posts into vectors for cosine search; a convolutional network extracts vectors for 1 million Wikimedia images, imported into BigQuery and processed as nearest‑neighbor joins; ML Engine with a GPU finishes the million‑image embedding job in about ten hours. Uses user-defined functions, feature vectors, and ML Engine GPU to run smart queries over petabytes.
AI-written summary · about
chapters (9)
- 0:00Introduction and speaker
- 1:57Finding similar pictures
- 3:22BigQuery capabilities
- 9:13Query execution infrastructure
- 9:54Smart analytics with feature vectors
- 15:13TensorFlow for pattern recognition
- 18:07Classifying Manhattan taxi trips
- 20:37Image similarity with vectors
- 24:13Summary and applications
related talks
47mBuilding intelligent analytics at scale
33mBringing Vector Search to OLTP
51mDiscover BigQuery ML, build your own CREATE MODEL stmt
27mUnified AI & Relational Power: Scalable Distributed Vector Search with
41mAnalyzing Terabytes of Data with Google BigQuery
12mVector Databases: Why QA professionals Needs to Care About them in the Age of AI?
45mForget SQL! Create your own AI Data Analyst
35mPowering Interactive Data Analysis with Google BigQuery