FoundationDB as Datastore for Training Deep Learning Models
Talkat FoundationDB Summit 2019 · on Monday, 2019-11-18· at 13:40· Grand Ballroom 5, Marriott Marquis San Diego Marina·
summary
Axstrm uses FoundationDB as the datastore and orchestration layer for a swim-tracking mobile app that stores extensive sensor and image data, spins out per-user isolated datasets, trains and validates per-device micro models, and ships updates back to devices. The pipeline starts from a shipped base model that hits ~80% baseline accuracy, collects manual user corrections, validates with a holdout set, retrains until a quality threshold, and runs in production with a low FoundationDB footprint.
AI-written summary · about
chapters (3)
related talks
46mFoundationDB: The Best-Kept Secret of New Distributed Architectures!
22mSolving Everyday Data Problems with FoundationDB
34mFoundationDB, the Black Knight
37mTechnical Overview of FoundationDB
22mHow FoundationDB powers SnowflakeDB's metadata!
50mBuilding the Next-Generation Backend on FoundationDB
26mRunning FDB at scale at Wavefront
25mFoundationDB Document Layer