Machine Learning / 01
ML Foundations
Build a foundation in learning systems, training, inference, and generalization.
Training finds a pattern. Generalization proves it was worth finding.
Here, generalization means evidence from appropriate held-out data. It is not a guarantee for every future input.
This is a knowledge index. Published titles open their articles; the remaining entries outline planned coverage.
Subtopics
- 01.01
What is Machine Learning?
Start with the gap between memorizing examples and learning a pattern that survives new data.
- 01.02
Why Machine Learning?
Decide when learning from examples is useful and when a simpler rule may be enough.
- 01.03
AI vs ML vs Deep Learning
Place machine learning and deep learning within the wider field of artificial intelligence.
- 01.04
Training vs Inference
Compare the work of fitting a model with using it to make predictions.
- 01.05
Model vs Algorithm
Distinguish the learned model from the procedure used to fit it.
- 01.06
Data, Features, Labels & Targets
Identify the inputs, observations, labels, and targets in a learning problem.
- 01.07
Supervised Learning
Frame a learning problem using examples with known targets.
- 01.08
Unsupervised Learning
Explore learning problems where examples do not come with target labels.
- 01.09
Semi-Supervised Learning
Explore how a learning setup can use both labeled and unlabeled examples.
- 01.10
Self-Supervised Learning
Explore how learning tasks can be constructed from the data itself.
- 01.11
Reinforcement Learning
Introduce learning through actions, feedback, and rewards.
- 01.12
Batch Learning
Plan training around a collected batch of examples.
- 01.13
Online Learning
Explore updates as new observations arrive and the checks those updates need.
- 01.14
Parametric Models
Examine models described by a fixed set of learned parameters.
- 01.15
Non-Parametric Models
Explore flexible model families and how their capacity relates to the data.
- 01.16
Hypothesis Space
Describe the candidate solutions a learning procedure can consider.
- 01.17
Inductive Bias
Identify the assumptions that guide learning beyond the observed examples.
- 01.18
Generalization
Ask whether a learned pattern holds on appropriate unseen examples.
- 01.19
Underfitting
Investigate when a model misses useful patterns in the training data.
- 01.20
Overfitting
Explore why strong training performance can coexist with weaker performance on unseen data.
- 01.21
Bias
Study the simplifying assumptions and systematic errors in a learning setup.
- 01.22
Variance
Examine how a fitted model can change across different training samples.
- 01.23
Bias-Variance Tradeoff
Compare sources of prediction error when choosing model complexity.
- 01.24
Model Complexity
Explore capacity, training fit, and evaluation beyond the training set.