Startup Showcase
summary
Four startups show concrete wins not vague promises: UnifyID claims passive, on‑device authentication deployed on 34 million devices with ~1 in 50,000 false‑positive motion-based accuracy and on‑device training to resist adversarial and privacy issues; InAccel runs FPGAs from Jupyter/Keras/Scikit with 10–20x speedup, zero code changes (ResNet50 at ~3,000 FPS, 7 ms latency on AWS F1); Blueshift uses XGBoost LambdaMART LTR with MAP@K gains and feature buckets (relevance, quality, fatigue, affinity); PerceptiLabs is a PyPI visual TensorFlow API that autogenerates code and per‑variable training visualizations.
AI-written summary · about
chapters (12)
- 0:00Start
- 2:31Implicit Authentication
- 19:58Pricing Model
- 24:43What Is the Need for Learning To Rank
- 25:56Feature Sets
- 26:28User Activity
- 30:21Martin Isaacson
- 31:21Challenges with Modeling and Debugging Models
- 32:18Visualizations
- 32:59Preceptor Labs
- 33:51Drag and Drop Interface
- 35:49Model Hub
related talks
01h 00mVCs vs Startups - Startup Pitch
47mAI Pitch: Beyond the Hype
40mMachines are Learning
16mEdge AI needs to be for everyone: breaking barriers to adoption
27m[Opening Keynote] Anyscale Demo: Machine Learning Application from Dev to Prod
01h 00mOpening Keynote
17mAI/ML demo derby
12mRevolutionizing Predictive Analytics