Scribd
Upload
23 views1 page

Introduction to Unsupervised Learning

The document discusses the differences between supervised and unsupervised learning, highlighting that supervised learning relies on labeled data for training, while unsupervised learning identifies similarities and clusters data without known outputs. It emphasizes the challenges of obtaining labeled targets and the use of internal measures for unsupervised learning. Additionally, it introduces the k-Means algorithm as a method for clustering similar inputs.

Uploaded by

Josh Minaj
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
23 views1 page

Introduction to Unsupervised Learning

The document discusses the differences between supervised and unsupervised learning, highlighting that supervised learning relies on labeled data for training, while unsupervised learning identifies similarities and clusters data without known outputs. It emphasizes the challenges of obtaining labeled targets and the use of internal measures for unsupervised learning. Additionally, it introduces the k-Means algorithm as a method for clustering similar inputs.

Uploaded by

Josh Minaj
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.

Supervised Learning relies on labeled target data or scoring systems to


indicate prediction quality.
2. Targets are helpful for training algorithms by showing the correct output
for given inputs.
3. Obtaining targets can be difficult, often requiring manual labeling, and
is not always biologically plausible.
4. Unsupervised Learning operates without any information about correct
outputs.
5. Unsupervised learning aims to find similarities between inputs and
cluster similar data together.
6. Classification in unsupervised learning is performed implicitly by
grouping similar inputs without knowing the true classes.
7. Clustering helps in identifying input groups that could correspond to
different classes without explicit labels.
8. Dimensionality reduction, covered earlier, is an example of unsupervised
learning focused on clustering similar datapoints.
9. Supervised learning minimizes external error criteria (like sum-of-squares
error) based on known targets.
10. Unsupervised learning cannot use target-based error criteria, requiring
internal, task-independent measures.
11. The error measure in unsupervised learning must not rely on external
information or be task-specific.
12. To guide learning, the concept of input space and weight space is
used.
13. Similarity between inputs is determined by how close their vectors are
using measures like Euclidean distance.
14. Clustering is based on grouping inputs that are close in input space and
not clustering those far apart.
15. Weight vectors in a network are aligned with input vectors to find
good matches and group similar inputs.
16. k-Means Algorithm is introduced as a simple and longstanding clustering
method in statistics.

You might also like