MLRG/fall10
From ResearchWiki
(Difference between revisions)
m (→Schedule) |
m (→Schedule) |
||
| (3 intermediate revisions not shown) | |||
| Line 48: | Line 48: | ||
| colspan="4" bgcolor="#dddddd" style = "text-align:center" | '''Active Learning''' | | colspan="4" bgcolor="#dddddd" style = "text-align:center" | '''Active Learning''' | ||
|- | |- | ||
| - | | Nov 12 || Pool-based active learning, Query by committee, Query by uncertainty || [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.89.4888&rep=rep1&type=pdf Support Vector Machine Active Learning with Applications to Text Classification], [http://pages.cs.wisc.edu/~bsettles/pub/settles.activelearning.pdf Active Learning survey (sections 3.1 and 3.2)] || [http://www.cs.utah.edu/~ngilbert Nathan] | + | | Nov 12 || Pool-based active learning, Query by committee, Query by uncertainty || [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.89.4888&rep=rep1&type=pdf Support Vector Machine Active Learning with Applications to Text Classification], [http://pages.cs.wisc.edu/~bsettles/pub/settles.activelearning.pdf Active Learning survey (sections 3.1 and 3.2)] || <s>[http://www.cs.utah.edu/~ngilbert Nathan]</s> |
|- | |- | ||
| - | | Nov 19 || Stream-based active learning || [http://www.jmlr.org/papers/volume7/cesa-bianchi06b/cesa-bianchi06b.pdf Worst-Case Analysis of Selective Sampling for Linear Classification] || | + | | Nov 19 || Pool-based active learning, Query by committee, Query by uncertainty, Stream-based active learning || [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.89.4888&rep=rep1&type=pdf Support Vector Machine Active Learning with Applications to Text Classification], [http://pages.cs.wisc.edu/~bsettles/pub/settles.activelearning.pdf Active Learning survey (sections 3.1 and 3.2)], [http://www.jmlr.org/papers/volume7/cesa-bianchi06b/cesa-bianchi06b.pdf Worst-Case Analysis of Selective Sampling for Linear Classification] || [http://www.cs.utah.edu/~ngilbert Nathan] |
|- | |- | ||
| *Nov 26 || Dealing with sampling bias and using cluster-structure for active learning || [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.149.5701&rep=rep1&type=pdf Hierarchical Sampling for Active Learning] || | | *Nov 26 || Dealing with sampling bias and using cluster-structure for active learning || [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.149.5701&rep=rep1&type=pdf Hierarchical Sampling for Active Learning] || | ||
|- | |- | ||
| - | | Dec 3 || Semi-supervised learning and active learning || [http://pages.cs.wisc.edu/~jerryzhu/pub/zglactive.pdf Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions] || | + | | Dec 3 || Semi-supervised learning and active learning || [http://pages.cs.wisc.edu/~jerryzhu/pub/zglactive.pdf Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions] || Piyush |
|- | |- | ||
| - | | Dec 10 || Multiview active learning || [http://www.isi.edu/~muslea/PS/muslea06a.pdf Active Learning with Multiple Views] || | + | | Dec 10 || Multiview active learning || [http://www.isi.edu/~muslea/PS/muslea06a.pdf Active Learning with Multiple Views] || (To be rescheduled) |
|} | |} | ||
| Line 68: | Line 68: | ||
Other papers: | Other papers: | ||
* On Co-Training: [http://books.nips.cc/papers/files/nips17/NIPS2004_0399.pdf Co-Training and Expansion: Towards Bridging Theory and Practice], [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.144.3460&rep=rep1&type=pdf Bayesian Co-training], [http://www.icml2010.org/papers/275.pdf A New Analysis of Co-Training] | * On Co-Training: [http://books.nips.cc/papers/files/nips17/NIPS2004_0399.pdf Co-Training and Expansion: Towards Bridging Theory and Practice], [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.144.3460&rep=rep1&type=pdf Bayesian Co-training], [http://www.icml2010.org/papers/275.pdf A New Analysis of Co-Training] | ||
| + | * [http://cseweb.ucsd.edu/~dasgupta/papers/twoface.pdf Two faces of active learning] | ||
Will be updated with more papers. | Will be updated with more papers. | ||
Latest revision as of 22:33, 3 December 2010
Semisupervised and Active Learning
Fri 2:00-3:20pm
MEB 3105
Contents |
Synopsis
Supervised learning algorithms usually require a good amount of labeled data in order to learn a reliable model. Since getting large quantities of labeled data can be expensive and/or difficult, much effort in machine learning has been devoted on coming up with ways to learn with a limited amount of labeled data. There are many ways of doing this. Two very important paradigms we will be looking at in this seminar are (1) semi-supervised learning which involves augmenting a small amount of available labeled data with a large amount of additional unlabeled data (which is usually very easy to obtain), and (2) active learning which involves judiciously selecting the most informative/useful labeled examples to be given to a supervised learning algorithm. In this seminar, we will be looking at some representative papers from both these paradigms. As it will not be possible to cover all important papers in a single seminar, for those interested, a bunch of papers will be added under the suggested readings.
Participants
- Piyush Rai, PhD Student, School of Computing
- Suresh Venkat, Asst. Prof, School of Computing
- Ruihong Huang, PhD Student, School of Computing
- Lalindra De Silva, PhD Student, School of Computing
- Sandeep P, MS Student, School of Computing
- Neal Richter, PhD (Montana State U), Working at the Rubicon Project
- Nathan, PhD Student, School of Computing
Schedule
(subject to change; * means will probably need a rescheduling)
| Date | Topic | Outline and Paper(s) | Presenter |
|---|---|---|---|
| Sep 3 | Outline, Motivation | Seminar logistics. Introduction to semisupervised learning and active learning | Piyush |
| Semisupervised Learning | |||
| Sep 10 | Bootstrapping/weak-supervision | Combining Labeled and Unlabeled Data with Co-Training (for some theoretical results, also see PAC Generalization Bounds for Co-training) | Piyush |
| Sep 17 | Low density regions and the cluster assumption for SSL | Semi-Supervised Classification by Low Density Separation, (also see section 5 of the SSL survey for other methods and further references) | Ruihong |
| Sep 24 | Imposing function smoothness: Graph based SSL | Manifold Regularization for Semi-supervised Learning, (also see section 6 of the SSL survey for other methods and further references) | Sandeep |
| Oct 1 | Probabilistic approaches: Expectation Maximization for SSL | Semi-Supervised Text Classification Using EM | |
| Oct 8 | Using unlabeled data to learn predictive functional structures | A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data | Sandeep |
| Oct 22 | Semi-supervised Learning for Ranking | Learning to Rank with Partially-Labeled Data | Neal |
| Oct 29 | | A PAC-style Model for Learning from Labeled and Unlabeled Data CANCELED | |
| Nov 5 | Semi-unsupervised Learning (for Clustering/Dimensionality Reduction) | Integrating constraints and metric learning in semi-supervised clustering, Semi-Supervised Dimensionality Reduction | Ruihong |
| Active Learning | |||
| Nov 12 | Pool-based active learning, Query by committee, Query by uncertainty | Support Vector Machine Active Learning with Applications to Text Classification, Active Learning survey (sections 3.1 and 3.2) | |
| Nov 19 | Pool-based active learning, Query by committee, Query by uncertainty, Stream-based active learning | Support Vector Machine Active Learning with Applications to Text Classification, Active Learning survey (sections 3.1 and 3.2), Worst-Case Analysis of Selective Sampling for Linear Classification | Nathan |
| *Nov 26 | Dealing with sampling bias and using cluster-structure for active learning | Hierarchical Sampling for Active Learning | |
| Dec 3 | Semi-supervised learning and active learning | Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions | Piyush |
| Dec 10 | Multiview active learning | Active Learning with Multiple Views | (To be rescheduled) |
Suggested Readings
Some survey papers:
- Semisupervised Learning Literature Survey
- Learning with Labeled and Unlabeled Data
- Active Learning Literature Survey
Other papers:
- On Co-Training: Co-Training and Expansion: Towards Bridging Theory and Practice, Bayesian Co-training, A New Analysis of Co-Training
- Two faces of active learning
Will be updated with more papers.
