Stages of ML
Training → Inference
Training
learning new capabilities by comparing data features to output labels
Features
the input used as comparison to the output label
Inference
using the now trained model to infer new predictions
Supervised ML (LABELS,FEATURES) (HAS NONE,ONE,BOTH)
both
Unsupervised ML (LABELS,FEATURES) (HAS NONE,ONE,BOTH)
does not have labels for its features
Reinforcement ML (LABELS,FEATURES) (HAS NONE,ONE,BOTH)
NONE
Types of Supervised Learning based on being…
Continuous or Categorical
Continuous
expressed in a range of values
Categorical
expressed under a category/name
Continuous Use Cases
House Prices, Temperature, Height
Regression
learning relationships between input features and target variables
Binary Classification
categorizes data into one of two classes (yes,no) (blue,red) (0,1)
Multi-Class Classification
categorizes data into one of three or more distinct categories
Logistic Regression
predicts if something is true/false; Binary; Uses a Sigmoid Function to do so
Linear Regression
Continuous; Regression; takes the aggregated values, draws a plane(line) that can be used to best predict information
Reinforcement Learning Components (Ag, Env, Sta, Ac, Po)
Agent, Environment, State, Action, Policy
Reinforcement Learning Agent
interacts w/ environment, takes actions, learns from feedback
Reinforcement Learning Environment
system where the agent reacts
Reinforcement Learning State
represents the current situation of the environment at a time
Reinforcement Learning Action
possible moves/decisions the agent can do in a state
Reinforcement Learning Policy
mapping the commands the agent uses to decide what action to take
Classification
for predicting a discrete label/category for each input
Uni-variate Linear Regression (Simple Linear Regression)
using only one feature to get the correct label
