Robots, houses & the IoT - Microsoft 's Brandon Rohrer on making things intelligent
summary
Claim: putting a learning brain between existing sensors and actuators turns a house into an adaptive robot rather than a blunt instrument. Example: the thermostat example senses temperature and actuates heat but needs learning to handle exceptions and preferences. Rule lists fail because the world surprises them, so methods like deep learning and specifically deep reinforcement learning are the tools proposed. Domains with traction are cars near production and manufacturing where sensors, control, and ROI make learning practical.
AI-written summary · about
related talks
13mReinforcement learning and it's growing role in AI
41mArtificial Intelligence: Let's Think About Learning
06mRL in real life: Bringing reinforcement learning to the enterprise
35mIndustrial Applications of Reinforcement Learning
18mRobotics: why now?
12mMachine Learning for Adaptive, Self-Optimizing Systems
01h 35mDeep Robotic Learning
04mHuman-AI interaction: Autonomous service robots