semisupervised learning a brief introduction. semisupervised learning introduction types of...

36
Semisupervis ed Learning A brief introduction

Upload: rosalyn-greene

Post on 30-Dec-2015

242 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning

A brief introduction

Page 2: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning

• Introduction• Types of semisupervised learning• Paper for review• References

Page 3: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Introduction

This is an extension to supervised learning. We have two sets of data:

Motivation: labeled data is sometimes hard to obtain.

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 4: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

An example from Mars Data Analysis

Digital Elevation MapGeomorphic Map

Martian landscape

Manually drawn geomorphic map of this landscape

Geomorphic map shows landforms chosen and defined by a domain expert.

Page 5: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Segmentation

Page 6: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Segmentation: Results.

Displayed on an elevation background.

2631 segments homogeneous in slope, curvature and flood.

Page 7: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Classification: Labeling.

A representative subset of objects are labeled as one of the following six classes:o Plaino Crater Flooro Convex Crater Wallso Concave Crater Wallso Convex Ridgeso Concave Ridges

Labeled segments.

Page 8: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Introduction

How can we learn from unlabeled data at all?

The answer lies in the set of assumptions about theunlabeled data distribution.

If assumptions are right, an advantage can be obtainedusing unlabeled data

But a decrease in performance is possible if assumptions are incorrect.

Page 9: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Introduction

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 10: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Introduction

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 11: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning

• Introduction• Types of semisupervised learning• Paper for review• References

Page 12: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Types of semi-supervised learning:

• Self-Training

• Generative Models

• Graph-Based Algorithms

• Multi-View Algorithms

• SVMs

Page 13: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Self-Training

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 14: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Self-Training Variations

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 15: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 16: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 17: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 18: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 19: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 20: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 21: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 22: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Graph-Based Models

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 23: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: TypesGraph-Based Models

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 24: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Multi-View Algorithms

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 25: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Multi-View Algorithms

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 26: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

SVMs

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 27: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

Figure obtained from X. Zhu. Semi-Supervised Learning Tutorial. ICML 2007

Page 28: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Types

A comparison of different approaches:

Self-Training: Method is simple, but early mistakes propagate down and can be very harmful.

Generative Models: great method if prob. model is correct, but it isdifficult to verify model correctness. In fact, unlabeled data may yielda decrease in accuracy if model is wrong.

Graph-Based Models: good solid mathematical solution if a graph is a good representative of the data distribution.

Multi-view Method: simple method that is less sensitive to errors inclassification, but there may not be a natural split of the features. SVMs: it can be used wherever SVM is applicable; but may fall into a local maxima, and optimization is hard.

Page 29: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning

• Introduction• Types of semisupervised learning• Paper for review• References

Page 30: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Paper

Unlabeled data: Now it helps, now it doesn’t by Singh, et. al.

Problem: analyze when semi-supervised learning helps to improve generalizationperformance.

Figure obtained from: Singh, et. al. Unlabeled data: now it helps, now it doesn’t. NIPS (2008).

Page 31: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Paper

Unlabeled data: Now it helps, now it doesn’t by Singh, et. al.

Some terminology that is necessary for the paper follows:

“The cluster assumption” means that the distributions of classes in the feature space is smooth on each set D ∈ D. The sets in D are called decision sets.

Figure obtained from: Singh, et. al. Unlabeled data: now it helps, now it doesn’t. NIPS (2008).

Page 32: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Paper

Unlabeled data: Now it helps, now it doesn’t by Singh, et. al.

Figure obtained from: Singh, et. al. Unlabeled data: now it helps, now it doesn’t. NIPS (2008).

Page 33: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Paper

Unlabeled data: Now it helps, now it doesn’t by Singh, et. al.

Text obtained from: Singh, et. al. Unlabeled data: now it helps, now it doesn’t. NIPS (2008).

Main Result: “…if the sets D are discernible using unlabeled data (the margin is large enough compared to average spacing between unlabeled data points), then there exists a semi-supervised learner that can perform as well as a supervised learner with clairvoyant knowledge of the decision sets, provided m ≫ n…”

Page 34: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning: Paper

Unlabeled data: Now it helps, now it doesn’t by Singh, et. al.

Figure obtained from: Singh, et. al. Unlabeled data: now it helps, now it doesn’t. NIPS (2008).

Page 35: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning

• Introduction• Types of semisupervised learning• Paper for review• References

Page 36: Semisupervised Learning A brief introduction. Semisupervised Learning Introduction Types of semisupervised learning Paper for review References

Semisupervised Learning