📚
课程 3 · 年级 8+
Three ways of learning
📖 课程
──▶
🎯 测验——选择翻译
──▶
✅ 本课已完成
Machines learn three ways. Supervised — "with a teacher": examples with answers (photo → "cat/dog"). Unsupervised — "without a teacher": the system groups similar things itself (customers by interests). Reinforcement — "by experience": an agent tries actions and gets rewards — that is how AlphaGo beat the Go world champion.
- ●Supervised: speech recognition, image diagnosis
- ●Unsupervised: recommendations, clusters, anomalies
- ●Reinforcement: games, robots, traffic control
🎯 Data beats algorithm: bad data → bad AI. Collecting and labeling data is its own profession.