Multiple instance learning for sparse positive bags

By Razvan C. Bunescu, Raymond J. Mooney. 2007

222 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Video Representation Learning Using Discriminative Pooling (2018, background): “Mathematically, suppose the tuple (wi, bi) represents the parameters of a max-margin hyperplane separating some of the features in a positive bag X i from all features in X, then we cast the following objective, which is a variant of the sparse MIL (SMIL) [3], normalized set kernel (NSK) [15], and ∝”
Multiple instance learning for soft bags via top instances (2015, methodology): “The positive source is the distribution of the target concept (e.g., image patches of “flower”), the negative source the distribution of background clutter (e.g., image patches of everything else).”
Partially Does It: Towards Scene-Level FG-SBIR with Partial Input (2022, unknown): “Local-MIL on the other hand, utilise multiple instance learning (MIL) paradigm [13] where the minimum cosine distance pair between the set of localised sketch and photo is considered in loss computation.”
Keywords to visual categories: Multiple-instance learning forweakly supervised object categorization (2008, background): “To recover a discriminative classifier between positive and negative bags of images, we consider the objective function suggested in [5] to determine a large-margin decision boundary while accounting for the fact that positive bags can be arbitrarily sparse.”
Multi-Level Active Prediction of Useful Image Annotations for Recognition (2008, methodology): “[26, 24, 23]).”
From Group to Individual Labels Using Deep Features (2015, methodology): “Other approaches include trans-ductive SVMs [4] and approaches for sparse multi-instance learning [3], as well as methods for identifying key instances and connect them through labels, which is very useful for discovering regions of interest in images [17, 18].”
Multi-instance learning by treating instances as non-I.I.D. samples (2008, unknown): “84.3 DD-SVM [10] 85.8 91.3 MissSVM [41] 87.6 80.0 MILES [9] 86.3 87.7 Diverse Density [22] 88.9 82.5 RELIC [27] 83.7 87.3 MITI [6] 83.7 88.2 Citation-kNN [34] 92.4 86.3 RIPPER-MI [12] 88.0 77.0 sbMIL [8] 91.8 87.7 MIBoosting [36] 87.9 84.0 MILBoosting [4] 92.0 87.1 MI-LR [25] 86.7 87.0 MULTINST [3] ”
Confidence-Rated Multiple Instance Boosting for Object Detection (2014, background): “The classifier ϕ is trained to achieve a low empirical error rate on the bags such that max j ( 2 · 1{ϕ(xij)≥0} − 1 ) = yi.”
Multiple Instance Learning: Algorithms and Applications (2008, methodology): “Instead, Bunescu et al . propose a third alternative for adapting an SVM to the MIL problem [11].”
Multiple instance classification: Review, taxonomy and comparative study (2013, methodology): “We refer to [20] for other SVM-based IS classiﬁers, being SbMIL the one with highest performance according to reported results [20].”
Multiple instance learning: A survey of problem characteristics and applications (2016, methodology): “For example, the scenario where the number of positive instances in a bag is low was referred to as either sparse bags [26, 27] or low witness rate [24, 28].”
Maximum Margin Multiple Instance Clustering With Applications to Image and Text Clustering (2011, background): “In [23], the authors make some modifications on the loss functions of single-instance SVM and focus more on the positive bags with smaller sizes.”