Results 1 
9 of
9
Towards Discovering What Patterns Trigger What Labels ∗
"... In many real applications, especially those involving data objects with complicated semantics, it is generally desirable to discover the relation between patterns in the input space and labels corresponding to different semantics in the output space. This task becomes feasible with MIML (MultiInst ..."
Abstract

Cited by 13 (8 self)
 Add to MetaCart
(Show Context)
In many real applications, especially those involving data objects with complicated semantics, it is generally desirable to discover the relation between patterns in the input space and labels corresponding to different semantics in the output space. This task becomes feasible with MIML (MultiInstance MultiLabel learning), a recently developed learning framework, where each data object is represented by multiple instances and is allowed to be associated with multiple labels simultaneously. In this paper, we propose KISAR, an MIML algorithm that is able to discover what instances trigger what labels. By considering the fact that highly relevant labels usually share some patterns, we develop a convex optimization formulation and provide an alternating optimization solution. Experiments show that KISAR is able to discover reasonable relations between input patterns and output labels, and achieves performances that are highly competitive with many stateoftheart MIML algorithms.
Multiinstance multilabel learning with weak label
 in Proceedings of the 23rd International Joint Conference on Artificial Intelligence
"... MultiInstance MultiLabel learning (MIML) deals with data objects that are represented by a bag of instances and associated with a set of class labels simultaneously. Previous studies typically assume that for every training example, all positive labels are tagged whereas the untagged labels are al ..."
Abstract

Cited by 6 (3 self)
 Add to MetaCart
MultiInstance MultiLabel learning (MIML) deals with data objects that are represented by a bag of instances and associated with a set of class labels simultaneously. Previous studies typically assume that for every training example, all positive labels are tagged whereas the untagged labels are all negative. In many real applications such as image annotation, however, the learning problem often suffers from weak label; that is, users usually tag only a part of positive labels, and the untagged labels are not necessarily negative. In this paper, we propose the MIMLwel approach which works by assuming that highly relevant labels share some common instances, and the underlying class means of bags for each label are with a large margin. Experiments validate the effectiveness of MIMLwel in handling the weak label problem. 1
2013b. Fast multiinstance multilabel learning
"... In multiinstance multilabel learning (MIML), one object is represented by multiple instances and simultaneously associated with multiple labels. Existing MIML approaches have been found useful in many applications; however, most of them can only handle moderatesized data. To efficiently handle ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
(Show Context)
In multiinstance multilabel learning (MIML), one object is represented by multiple instances and simultaneously associated with multiple labels. Existing MIML approaches have been found useful in many applications; however, most of them can only handle moderatesized data. To efficiently handle large data sets, we propose the MIMLfast approach, which first constructs a lowdimensional subspace shared by all labels, and then trains label specific linear models to optimize approximated ranking loss via stochastic gradient descent. Although the MIML problem is complicated, MIMLfast is able to achieve excellent performance by exploiting label relations with shared space and discovering subconcepts for complicated labels. Experiments show that the performance of MIMLfast is highly competitive to stateoftheart techniques, whereas its time cost is much less; particularly, on a data set with 30K bags and 270K instances, where none of existing approaches can return results in 24 hours, MIMLfast takes only 12 minutes. Moreover, our approach is able to identify the most representative instance for each label, and thus providing a chance to understand the relation between input patterns and output semantics.
Efficient MaxMargin MultiLabel Classification with . . .
 MACH LEARN
"... The goal in multilabel classification is to tag a data point with the subset of relevant labels from a prespecified set. Given a set of L labels, a data point can be tagged with any of the 2 L possible subsets. The main challenge therefore lies in optimisingover thisexponentially largelabel space ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
The goal in multilabel classification is to tag a data point with the subset of relevant labels from a prespecified set. Given a set of L labels, a data point can be tagged with any of the 2 L possible subsets. The main challenge therefore lies in optimisingover thisexponentially largelabel space subject tolabel correlations. Our objective, in this paper, is to design efficient algorithms for multilabel classification when the labels are densely correlated. In particular, we are interested in the zeroshot learning scenario where the label correlations on the training set might be significantly different fromthose onthe testset. We propose a maxmargin formulation where we model prior label correlations but do not incorporate pairwise label interaction terms in the prediction function. We show that the problem complexity can be reduced from exponential to linear while modelling dense pairwise prior label correlations. By incorporating relevant correlation priors we can handle mismatches between the training and test set statistics. Our proposed formulation generalises the effective 1vsAll method and we provide aprincipled interpretation ofthe 1vsAlltechnique. We develop efficient optimisation algorithms for our proposed formulation. We adapt the Sequential Minimal Optimisation (SMO) algorithm to multilabel classification and show that, with some bookkeeping, we can reduce the training time from being superquadratic to almost linear in the number of labels. Furthermore, by effectively reutilizing the kernel cache and jointly optimising over all variables, we can be orders of magnitude faster than the competing stateoftheart algorithms. We
unknown title
, 2013
"... This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal noncommercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or sel ..."
Abstract
 Add to MetaCart
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal noncommercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit:
1 Improvement of Learning Algorithm for the Multiinstance Multilabel RBF Neural Networks Trained with Imbalanced Samples
"... Multiinstance multilabel learning (MIML) is a novel learning framework where each sample is represented by multiple instances and associated with multiple class labels. In several learning situations, the multiinstance multilabel RBF neural networks (MIMLRBF) can exploit connections between the ..."
Abstract
 Add to MetaCart
(Show Context)
Multiinstance multilabel learning (MIML) is a novel learning framework where each sample is represented by multiple instances and associated with multiple class labels. In several learning situations, the multiinstance multilabel RBF neural networks (MIMLRBF) can exploit connections between the instances and the labels of an MIML example directly. However, it is quite often that the numbers of samples in different categories are discrete, i.e., the class distribution is imbalanced. When an MIMLRBF is trained with imbalanced samples, it will produce poor performance for setting the consistent fraction parameter α for all classes. This paper presents an improved approach in learning algorithms used for training MIMLRBF with imbalanced samples. In the first cluster stage, the methodology calculates the initial medoids for each category based on the data density. Afterwards, kmedoids is been invoked to optimize the medoids. The network will take advantage of the welladjusted units. In the second stage, the weights between the first and second layer are optimized by the singular value decomposition method. The improved approaches could be used in applications with imbalanced samples. Comparing results employing diverse learning strategies shows interesting outcomes as have come out of this paper.
Proceedings of the TwentyThird International Joint Conference on Artificial Intelligence ScalingUp Security Games with Boundedly Rational Adversaries: A CuttingPlane Approach
"... To improve the current realworld deployments of Stackelberg security games (SSGs), it is critical now to efficiently incorporate models of adversary bounded rationality in largescale SSGs. Unfortunately, previously proposed branchandprice approaches fail to scaleup given the nonconvexity of su ..."
Abstract
 Add to MetaCart
To improve the current realworld deployments of Stackelberg security games (SSGs), it is critical now to efficiently incorporate models of adversary bounded rationality in largescale SSGs. Unfortunately, previously proposed branchandprice approaches fail to scaleup given the nonconvexity of such models, as we show with a realization called COCOMO. Therefore, we next present a novel cuttingplane algorithm called BLADE to scaleup SSGs with complex adversary models,with three key novelties: (i) an efficient scalable separation oracle to generate deep cuts; (ii) a heuristic that uses gradient to further improve the cuts; (iii) techniques for qualityefficiency tradeoff. 1
Evaluation of Joint MultiInstance MultiLabel Learning For Breast Cancer Diagnosis
"... Abstract—Multiinstance multilabel (MIML) learning is a challenging problem in many aspects. Such learning approaches might be useful for many medical diagnosis applications including breast cancer detection and classification. In this study subset of digiPATH dataset (whole slide digital breast ca ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract—Multiinstance multilabel (MIML) learning is a challenging problem in many aspects. Such learning approaches might be useful for many medical diagnosis applications including breast cancer detection and classification. In this study subset of digiPATH dataset (whole slide digital breast cancer histopathology images) are used for training and evaluation of six stateoftheart MIML methods. At the end, performance comparison of these approaches are given by means of effective evaluation metrics. It is shown that MIMLkNN achieve the best performance that is %65.3 average precision, where most of other methods attain acceptable results as well. I.
SEMISUPERVISED LEARNING WITH PARTIALLY LABELED EXAMPLES
, 2010
"... Traditionally, machine learning community has been focused on supervised learning where the source of learning is fully labeled examples including both input features and corresponding output labels. As one way to alleviate the costly effort of collecting fully labeled examples, semisupervised lea ..."
Abstract
 Add to MetaCart
Traditionally, machine learning community has been focused on supervised learning where the source of learning is fully labeled examples including both input features and corresponding output labels. As one way to alleviate the costly effort of collecting fully labeled examples, semisupervised learning usually concentrates on utilizing a large amount of unlabeled examples together with a relatively small number of fully labeled examples to build better classifiers. Even though many semisupervised learning algorithms are able to take advantage of unlabeled examples, there is a significant amount of effort in designing good models, features, kernels, and similarity functions. In this dissertation, we focus on semisupervised learning with partially labeled examples. Partially labeled data can be viewed as a tradeoff between fully labeled data and unlabeled data, which can provide additional discriminative information in comparison to unlabeled data and requires less human effort to collect than fully labeled data. In our setting of semisupervised learning with partially labeled examples, the learning method is provided with a large amount of partially labeled examples and is usually augmented with a relatively small set of fully labeled examples. Our main goal is to integrate partially labeled examples into the conventional learning framework, i.e. to build a more accurate classifier. The dissertation addresses four different semisupervised learning problems in presence of partially labeled examples. In addition, we summarize general principles for the semisupervised learning with partially labeled examples.