Đại học Kinh tế Quốc dân – National Economics UniversityMICRO4ASIA
🏆 Arena
Sign inSign up
MICRO4ASIAThe six micro-credentials

Courses are delivered on the Limio learning platform. Limio

← Fundamental of Machine Learning

You are viewing a public preview

Sign in to save your progress, do assignments, join discussions and get feedback.

Sign in
Module 1: Data Preparation and Model Foundations

1.3 Labeled Data and the Three Groups of Algorithms

Print / Save as PDF

1.3 Labeled Data and the Three Groups of Algorithms

After this unit you will: be able to distinguish labeled from unlabeled data, and name the three groups of machine learning algorithms.

Key Points

  • Unlabeled data is raw data; an annotator assigns labels (for example cat, rabbit, sport) to turn it into labeled data.
  • The three groups of algorithms: Supervised, Unsupervised and Reinforcement Learning.
  • Supervised Learning is illustrated by a picture of a teacher pointing out the answers (labeled data).
  • Unsupervised Learning works with unlabeled data, finding structure such as clusters.
  • Reinforcement Learning: an agent interacts with an environment through state, action and reward.

Study notes

By the end of this lesson you can:

  • Explain the difference between unlabeled and labeled data, and the role of an annotator.
  • Explain why labeling is often the most expensive part of a machine learning project.
  • Name the three families of machine learning algorithms: supervised, unsupervised and reinforcement learning.
  • Decide which family fits a given problem based on the data available.

Key idea: Whether your data has labels decides which kind of learning you can use. With labeled data (the correct answer is known) you can use supervised learning, the focus of this course; without labels you need unsupervised learning.

Unlabeled data and labeled data

The lecture shows three items in yellow: a drawing of a cat, a drawing of a rabbit and a news headline. They are unlabeled data: raw information with no extra context about what it is. A computer looking at the cat picture only sees pixels; it does not know it is a cat.

In the middle stands an annotator, usually a person. The annotator looks at each item and attaches the correct answer: this is a cat, this is a rabbit, this headline is about sport. The same items, now tagged and shown in green, are labeled data.

Unlabeled dataLabeled data
What it containsRaw inputs onlyInputs plus the correct answer (label)
ExampleA photo with no descriptionThe same photo tagged "cat"
Learning it allowsUnsupervised learningSupervised learning

Why the distinction matters

The type of data you have decides the type of algorithm you can use. If you know the correct answer for each example, you can train a model by comparing its predictions with those answers. If you do not, the model must find structure on its own.

Labeling is often expensive and slow, because humans usually have to go through and tag each example by hand. In real projects, collecting and labeling data can take far more time and money than training the model. It is one of the biggest practical bottlenecks in machine learning.

Three families of algorithms

The lecture shows three icons branching from "Machine learning algorithms":

FamilyIcon in the lectureDataHow it learnsExample use
Supervised learningA teacher at a blackboardLabeledCompares predictions with known correct answersPredicting house prices, spam detection
Unsupervised learningA colourful scatter of cluster nodesUnlabeledFinds hidden structure or patterns by itselfGrouping customers by purchasing behaviour
Reinforcement learningA small robot and its environmentRewards from interactionAn agent observes a state, takes an action, receives a reward or penaltyGame-playing AI, robot control

Supervised learning matches the teacher analogy from lesson 1.1: the model learns by checking its work against the solution. It is the most widely used and most foundational family, and almost all of this course is about it.

Unsupervised learning has no answer key. The algorithm tries to discover groups or patterns without being told what they should be.

Reinforcement learning is quite different. An agent interacts with an environment: it sees the current state, chooses an action and receives a reward or penalty. Over time it learns which actions lead to better rewards.

Worked example

A retail company has three ideas. Decide which family of algorithm each one needs.

  1. Predict whether a customer will cancel their membership next month. The company has two years of history and knows which past customers actually cancelled. The correct answer exists for every past example, so this is labeled data and calls for supervised learning.
  2. Find natural groups among 50,000 customers based on what they buy. Nobody has defined the groups in advance and there is no "correct group" column. The data is unlabeled, so this is unsupervised learning.
  3. Train software that sets discounts day by day and learns from the profit each choice produces. There is no fixed answer key; the system takes actions and receives a reward (profit) afterwards. This is the shape of reinforcement learning.

Now suppose the company wants to sort 10,000 product photos into "good" and "damaged" but none of the photos has a tag. Before any supervised model can be trained, someone must act as the annotator and label at least part of the photos. Budget time and money for that step first.

Common mistakes

  • Assuming any dataset can be used for supervised learning. Supervised learning needs a known correct answer for each training example.
  • Underestimating labeling cost. Tagging data by hand is often the slowest and most expensive part of the project.
  • Calling reinforcement learning "learning from labels". It learns from rewards received after actions, not from a list of correct answers.
  • Thinking unsupervised learning predicts a known target. It discovers structure; it does not predict a column you already have answers for.

Summary

  • Unlabeled data is raw input; labeled data also carries the correct answer added by an annotator.
  • Labels decide which kind of learning is possible.
  • Creating labels is costly and is a major bottleneck in real projects.
  • The three families are supervised, unsupervised and reinforcement learning.
  • This course focuses almost entirely on supervised learning.

Flashcards · 1.3 Labeled Data and the Three Groups of Algorithms

9 words
WordMeaningExample
  • Unlabeled data

    Raw data with no information about what each example is or which category it belongs to.

    A folder of animal photos with no tags or descriptions.

  • Labeled data

    Data where each example carries the correct answer, usually added by a human.

    A photo tagged "cat".

  • Annotator

    A person (or tool) who assigns the correct labels to raw data.

    Someone tagging a headline as "sport".

  • Supervised learning

    Learning from labeled data by comparing predictions with known correct answers.

    Predicting house prices from past sales with known prices.

  • Unsupervised learning

    Learning from unlabeled data by finding hidden structure or patterns without an answer key.

    Grouping customers by purchasing behaviour.

  • Reinforcement learning

    Learning by an agent that takes actions in an environment and receives rewards or penalties.

    A game-playing AI improving its moves.

  • Agent

    In reinforcement learning, the learner that observes a state and chooses actions.

    The robot in the lecture's diagram.

  • Reward

    The feedback signal an agent receives after an action, telling it how good the action was.

    +1 for winning a game, -1 for losing.

  • Labeling bottleneck

    The practical problem that labeling data by hand often costs more time and money than training the model.

    Tagging 10,000 production-line photos before training.