You can AI like an Expert
summary
Argues that a symbolic representation unifies images, code, nets and data so the system auto-selects encoders, methods (nearest neighbors, gradient boosted trees) and even loss layers, letting a single pipeline train, apply, inspect and deploy models. Shows a Titanic classifier with 77% accuracy, a rock‑paper‑scissors image classifier, a sine approximator, a net repository of ~100 shareable architectures and net‑surgery to swap output layers, and a hybrid pipeline that estimates vehicle speed at 20 m/s within 5 m by combining a CREP feature net with Doppler algebra.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:55Symbolic computation Examples
- 5:30Automated Machine Learnin The model workflow
- 6:41Building machine learning models
- 15:23Automating neural network
- 34:15Multiparadigm computation
related talks
48mKeynote – Sameer Singh
40mMachines are Learning
03mRethinking software engineering in the AI era
42mReasonable AI
46mCode + AI: Will Robots Take Our Coding Jobs? ML Applied to Programming
59mAI in Finance: Scope and Examples
41mKEYNOTE: Image processing, AI, and autonomous systems
41mModels Schm-odels: Why you Should Care about Data-Centric AI