Flashcards
Lecture 1: Machine Learning Basics
Flashcards do conjunto (14)
- 1Learning Problems: Supervised LearningSupervised Learning:
- Learn model parameters using dataset of data-label pairs {(xi, yi)}
Examples: Classification, regression, structured prediction - 2Learning Problem: Unsupervised LearningLearn model parameters using dataset without labels {xi}
Examples: Clustering, dimensionality reduction, generative models - 3Learning Problem: Self-supervised LearningLearn model parameters using dataset of data-data pars {xi, xi'}
Examples: Self-supervised stereo/flow, contrastive learning - 4Learning Problem: Reinforcement LearningLearn model parameters using active exploration from sparse rewards
Examples: deep q Learning, gradient policy - 5Supervised Learning: Classification/Regressioninputs x can be any kind of objects e.g. images, text, audio
Output y is a discrete or a real number e.g. classification, regression - 6Supervised Learning: Structured Output Learningf: X --> Y
Inputs X can be any kind of objects
outputs Y are complex structured objects like text, images, folds of a protein - 7Supervised LearningLearning: Estimate parameters w from data data {(xi, yi)}
Inference: Make novel Predictions y=fw(x) - 8Supervised Learning: Classification
- 9Supervised Learning: Regression
- 10Goal of Linear RegressionPredict y for a previously unseen input x
- 11Linear Regression: ErrorfunctionThe error function E(w) measures the displacement along the y dimension between the data points (green) and the model f(x,w) (red) specified with the parameters w
- 12The gradient of the error function wrt the parameters w is given by:
- 13Name some properties of E(w)As E(w) is quadratic and convex in w, its minimizer is given in closed form:
- 14—
Informações
- Descrição
- FlashCards sobre Lecture 1: Machine Learning Basics, criado por Nora Gerth em 22-03-2023.
- Criado por
- Nora Gerth
- Visualizações
- 2
- Avaliação