-
Google Speech API 퀵랩: 영어 음성 -> 영어 텍스트 리턴.
-
Google Natural language API 퀵랩: 자연어 텍스트에 대한 다양한 분석 결과를 내놓는 API이다. 예를 들어, 문장성분 분석/감정 분석/등등.
-
entire purpose of ML is to make predictions on new data.
-
you wil learn to differentiate b/w the major categories of ML problems.
-
identify why deep learning is so popular.
-
supervised learning
-
features -> label = prediction problem
-
regression = continuous labels. MSE.
-
Classification = discrete labels. make decision boundary. Cross-entropy.
-
-
unsupervised learning
- description problem
-
ML History
- linear regression: Y = W * X
- Perceptron: single layer perceptron was modeled on the human brain, but there are simple functions (XOR) that is cannot learn. ==>??
- Neural Network: nonlinear activation function 이 없다면 multi-layer perceptron은 single layer 로 압축된다.