Partially Supervised Classification

Liu, Zhigang; Shi, Wenzhong; Li, Deren; Qin, Qianqing

doi:10.4018/jdwm.2006070104

Cited by 29 publications

(56 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The positive class expansion problem appears to have some relationship with PULearning [12,17], concept drift [9,10], and covariate shift [8,1]. But in fact it is very different from these tasks.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

A framework for modeling positive class expansion with single snapshot

Zhou

2009

Knowl Inf Syst

View full text Add to dashboard Cite

Abstract. In many real-world data mining tasks, the coverage of the target concept may change as the time changes. For example,the coverage of "learned knowledge" of a student today may be different from his/er "learned knowledge" tomorrow, since the "learned knowledge" of the student is in expanding everyday. In order to learn a model capable of making accurate predictions, the evolution of the concept must be considered, and thus, a series of data sets collected at different time is needed. However, in many cases there is only a single data set instead of a series of data sets. In other words, only a single snapshot of the data along the time axis is available. In this paper, we show that for positive class expansion, i.e., the coverage of the target concept is in expanding as illustrated in the above "learned knowledge" example, we can learn an accurate model from the single snapshot data with the help of domain knowledge given by user. The effectiveness of the proposed framework is validated in experiments.

show abstract

Section: Related Workmentioning

confidence: 99%

“…A large part of works addressing PU-Learning follow two steps, e.g. [12,17]. First, strong negative instances are discovered from the unlabeled data.…”

Section: Related Workmentioning

confidence: 99%

A framework for modeling positive class expansion with single snapshot

Zhou

2009

Knowl Inf Syst

View full text Add to dashboard Cite

show abstract

“…PU learning has been investigated by several researchers (Liu et al, 2002;Denis et al, 2002;Li and Liu, 2003;Elkan and Noto, 2008;Hsieh et al, 2015). A popular approach follows a two-stage strategy: (i) identifying a set of reliable N class instances RN from the unlabeled set U ; and (ii) building a classifier using P (P class) and RN (N class) and Q = U − RN (unlabeled) by applying a learning algorithm (e.g., SVM) iteratively.…”

Section: Traditional Pu Learningmentioning

confidence: 99%

“…This is important because it gives us a formal model to tackle the problem. PU learning is stated as follows (Liu et al, 2002): given a set P of examples of a particular class (we also use P to denote the class) and a set U of unlabeled examples which contains hidden instances from both classes P and not-P (called N ), we want to build a classifier using P and U to classify the data in U or future test data into the two classes, i.e., P and N (or not-P). In our case, P is the existing sentiment lexicon, and U is a set of candidate words from a social media corpus.…”

Section: Introductionmentioning

confidence: 99%

Sentiment Lexicon Expansion Based on Neural PU Learning, Double Dictionary Lookup, and Polarity Association

Wang

Zhang

Liu

2017

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

Although many sentiment lexicons in different languages exist, most are not comprehensive. In a recent sentiment analysis application, we used a large Chinese sentiment lexicon and found that it missed a large number of sentiment words used in social media. This prompted us to make a new attempt to study sentiment lexicon expansion. This paper first formulates the problem as a PU learning problem. It then proposes a new PU learning method suitable for the problem based on a neural network. The results are further enhanced with a new dictionary lookup technique and a novel polarity classification algorithm. Experimental results show that the proposed approach greatly outperforms baseline methods.

show abstract

“…The objective in this situation is to perform standard machine learning activities despite this data restriction. The problem has been referred to in the literature under several names, including partially supervised classification [15], positive example based learning [19] and positive unlabelled learning [9]. A typical application has been text classification -given a number of query documents belonging to a particular class (e.g.…”

Section: Introductionmentioning

confidence: 99%

Statistical Hypothesis Testing in Positive Unlabelled Data

Sechidis

Calvo

Brown

2014

Machine Learning and Knowledge Discovery in Databases

View full text Add to dashboard Cite

Abstract. We propose a set of novel methodologies which enable valid statistical hypothesis testing when we have only positive and unlabelled (PU) examples. This type of problem, a special case of semi-supervised data, is common in text mining, bioinformatics, and computer vision. Focusing on a generalised likelihood ratio test, we have 3 key contributions:(1) a proof that assuming all unlabelled examples are negative cases is sufficient for independence testing, but not for power analysis activities; (2) a new methodology that compensates this and enables power analysis, allowing sample size determination for observing an effect with a desired power; and finally, (3) a new capability, supervision determination, which can determine a-priori the number of labelled examples the user must collect before being able to observe a desired statistical effect. Beyond general hypothesis testing, we suggest the tools will additionally be useful for information theoretic feature selection, and Bayesian Network structure learning.

show abstract

Partially Supervised Classification

Cited by 29 publications

References 18 publications

A framework for modeling positive class expansion with single snapshot

A framework for modeling positive class expansion with single snapshot

Sentiment Lexicon Expansion Based on Neural PU Learning, Double Dictionary Lookup, and Polarity Association

Statistical Hypothesis Testing in Positive Unlabelled Data

Contact Info

Product

Resources

About