Supervised versus multiple instance learning

Ray, Soumya; Craven, Mark

doi:10.1145/1102351.1102439

Cited by 182 publications

(156 citation statements)

References 12 publications

Supporting

Mentioning

151

Contrasting

Order By: Relevance

“…Multi-instance learning was originated from Dietterich et al (1997)'s research on drug activity prediction, since that it has been studied by a lot of researchers and many algorithms have been developed, such as Diverse Density (Maron & Lozano-Pérez, 1998) and EM-DD (Zhang & Goldman, 2002), the k-nearest neighbor algorithm Citation-kNN (Wang & Zucker, 2000), decision tree algorithms RELIC (Ruffo, 2000) and ID3-MI (Chevaleyre & Zucker, 2001), rule learning algorithm RIPPER-MI (Chevaleyre & Zucker, 2001), SVM algorithms MI-SVM and mi-SVM (Andrews et al, 2003) and DD-SVM (Chen & Wang, 2004), ensemble algorithms MI-Ensemble (Zhou & Zhang, 2003) and MIBoosting (Xu & Frank, 2004), logistic regression algorithm MI-LR (Ray & Craven, 2005), etc. Many of those algorithms were developed by adapting a single-instance supervised learning algorithm to multiinstance learning through shifting its focus from the discrimination on the instances to the discrimination on the bags (Zhou & Zhang, 2003).…”

Section: Multi-instance Learningmentioning

confidence: 99%

“…The results are tabulated in (Chen et al, 2006), MI-LR (Ray & Craven, 2005), MIBoosting (Xu & Frank, 2004), DD-SVM (Chen & Wang, 2004), mi-SVM and MI-SVM (Andrews et al, 2003), RIPPER-MI (Chevaleyre & Zucker, 2001), RELIC (Ruffo, 2000), Citation-kNN (Wang & Zucker, 2000), Diverse Density (Maron & Lozano-Pérez, 1998), MULTINST (Auer, 1997) and Iterated-discrim APR (Dietterich et al, 1997). Table 1 shows that on the Musk data MissSVM is competitive with state-of-the-art multi-instance learning algorithms.…”

Section: Drug Activity Predictionmentioning

confidence: 99%

See 1 more Smart Citation

On the relation between multi-instance learning and semi-supervised learning

Zhou

2007

Proceedings of the 24th International Conference on Machine Learning

138

View full text Add to dashboard Cite

Multi-instance learning and semi-supervised learning are different branches of machine learning. The former attempts to learn from a training set consists of labeled bags each containing many unlabeled instances; the latter tries to exploit abundant unlabeled instances when learning with a small number of labeled examples. In this paper, we establish a bridge between these two branches by showing that multi-instance learning can be viewed as a special case of semi-supervised learning. Based on this recognition, we propose the MissSVM algorithm which addresses multi-instance learning using a special semisupervised support vector machine. Experiments show that solving multi-instance problems from the view of semi-supervised learning is feasible, and the MissSVM algorithm is competitive with state-of-the-art multiinstance learning algorithms.

show abstract

Section: Multi-instance Learningmentioning

confidence: 99%

Section: Drug Activity Predictionmentioning

confidence: 99%

On the relation between multi-instance learning and semi-supervised learning

Zhou

2007

Proceedings of the 24th International Conference on Machine Learning

138

View full text Add to dashboard Cite

show abstract

“…for j = 1 : mi do 8: These two candidate values (i.e., (19) and (20)) are then compared, and the larger value is the solution of the lth subproblem in (16). With g features, there are thus a total of 2g candidates ford.…”

Section: Finding a Violated Constraintmentioning

confidence: 99%

“…The first is the Diverse Density (DD) algorithm [15] and its variants, e.g., EM-DD [26] and multi-instance logistic regression [19]. These methods apply gradient search with multiple restarts to identify an instance which maximizes the diverse density, that is, an instance close to every positive bags while far from negative bags.…”

Section: Introductionmentioning

confidence: 99%

A Convex Method for Locating Regions of Interest with Multi-instance Learning

Kwok

Tsang

et al. 2009

Machine Learning and Knowledge Discovery in Databases

View full text Add to dashboard Cite

Abstract. In content-based image retrieval (CBIR) and image screening, it is often desirable to locate the regions of interest (ROI) in the images automatically. This can be accomplished with multi-instance learning techniques by treating each image as a bag of instances (regions). Many SVM-based methods are successful in predicting the bag labels, however, few of them can locate the ROIs. Moreover, they are often based on either local search or an EM-style strategy, and may get stuck in local minima easily. In this paper, we propose two convex optimization methods which maximize the margin of concepts via key instance generation at the instance-level and bag-level, respectively. Our formulation can be solved efficiently with a cutting plane algorithm. Experiments show that the proposed methods can effectively locate ROIs, and they also achieve performances competitive with state-of-the-art algorithms on benchmark data sets.

show abstract

“…This is a scenario with latent variables in training, as the label of an individual instance in a bag is not observed; only the label of the whole bag. The model for β = 1 and no error terms recovers the MI/LR from [18], for β → ∞ the model reduces to the MI-SVM [19]. We construct a one-dimensional synthetic dataset which illustrates the deficiencies of the MI-SVM.…”

Section: Multiple Instance Learningmentioning

confidence: 99%

Entropy and Margin Maximization for Structured Output Learning

Pletscher

Ong

Buhmann

2010

Machine Learning and Knowledge Discovery in Databases

View full text Add to dashboard Cite

Abstract. We consider the problem of training discriminative structured output predictors, such as conditional random fields (CRFs) and structured support vector machines (SSVMs). A generalized loss function is introduced, which jointly maximizes the entropy and the margin of the solution. The CRF and SSVM emerge as special cases of our framework. The probabilistic interpretation of large margin methods reveals insights about margin and slack rescaling. Furthermore, we derive the corresponding extensions for latent variable models, in which training operates on partially observed outputs. Experimental results for multiclass, linear-chain models and multiple instance learning demonstrate that the generalized loss can improve accuracy of the resulting classifiers.

show abstract

Supervised versus multiple instance learning

Cited by 182 publications

References 12 publications

On the relation between multi-instance learning and semi-supervised learning

On the relation between multi-instance learning and semi-supervised learning

A Convex Method for Locating Regions of Interest with Multi-instance Learning

Entropy and Margin Maximization for Structured Output Learning

Contact Info

Product

Resources

About