Scribd
Upload
101 views2 pages

1.6 Assignment 1 - Introduction To Machine Learning

The document provides two definitions of machine learning: one by Tom Mitchell focusing on performance improvement through experience, and another by Arthur Samuel emphasizing the ability of computers to learn without explicit programming. It also discusses an application of machine learning in traffic light control, which adapts to real-time conditions, and contrasts supervised learning, which uses labeled data for tasks like income prediction, with unsupervised learning, which analyzes unlabeled data to find patterns, such as customer clustering. The references cite Shah (2024) from the University of Washington.

Uploaded by

kevincarlson45
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
101 views2 pages

1.6 Assignment 1 - Introduction To Machine Learning

The document provides two definitions of machine learning: one by Tom Mitchell focusing on performance improvement through experience, and another by Arthur Samuel emphasizing the ability of computers to learn without explicit programming. It also discusses an application of machine learning in traffic light control, which adapts to real-time conditions, and contrasts supervised learning, which uses labeled data for tasks like income prediction, with unsupervised learning, which analyzes unlabeled data to find patterns, such as customer clustering. The references cite Shah (2024) from the University of Washington.

Uploaded by

kevincarlson45
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

1. Provide two different definitions of machine learning.

Machine learning as defined by Tom Mitchell is “A computer program is said to learn from
experience E with respect to some class of tasks T and performance measure P if its performance
at tasks in T, as measured by P, improves with experience E. " (Shah, 2024). This definition
focuses on performance improvement through experience and connects learning to measurable
performance rather than broad knowledge.

Arthur Samuel defines machine learning as giving "computers the ability to learn without being
explicitly programmed" (Shah, 2024). This means computational systems can write their own
algorithms or change parameters for existing algorithms without human intervention.

2. Provide an example application where developing a machine learning approach is better


than implementing a deterministic algorithm.

Traffic light control at intersections is a good example because traditional deterministic programs
require humans to conduct traffic studies and manually adjust timing of traffic lights

A machine learning approach is better because it can use real-time camera data to automatically
adjust light timing patterns. The system learns optimal patterns by itself, adapting to changing
traffic conditions across different times of day and seasons without human intervention. This
results in more efficient traffic flow while reducing the need for costly human monitoring.

3. How do supervised learning and unsupervised learning differ? Give an example for each.

Supervised learning uses labeled data where we already know the "correct answers" to
learn patterns. Supervised learning problems "require us to know the truth first" (Shah,
2024). For example, income prediction based on age and occupation is a supervised
learning task. If we have a dataset with people's ages, occupations, and their actual
incomes, we can train a model to predict someone's income when given only their age
and occupation.

Unlike supervised learning, unsupervised learning works with unlabeled data and tries to
find patterns without knowing categories in advance. This is for situations "when the data
given to us do not have clear labels or true values. And yet, we are tasked with exploring
and explaining that data" (Shah, 2024). For example, in clustering approaches, an
algorithm might analyze purchase history and browsing behavior to group customers into
distinct clusters without being told what those groups should be.
References

Shah, C. (2024). Introduction to Machine Learning. University of Washington.

You might also like