Nonparametric Maximum Margin Similarity for Semi-Supervised Learning
نویسندگان
چکیده
1. Nonparametric Label Propagation (LP) has been proven to be effective for semi-supervised learning problems, and it predicts the labels for unlabeled data by a harmonic solution of an energy minimization problem which encourages local smoothness of the labels in accordance with the similarity graph. 2. On the other hand, the success of LP algorithms highly depends on the underlying similarity graph. Most similarity graphs for LP are constructed empirically and the objective function over the similarity graphs is defined as sum of the product of pairwise similarity and the squared label difference. 3. We relate LP to a novel nonparametric maximum margin similarity framework with the concept of similarity margin, and present a new semi-supervised learning algorithm called Maximum Margin Similarity Graph (MMSG). The conventional LP algorithm can be interpreted as a special case of our MMSG algorithm when the separation parameter is sufficiently large. 4. By the sample-based similarity margin rather than the expectation based margin, our framework leads to an tractable optimization problem which is solved by the projected subgradient method.
منابع مشابه
Minimum Density Hyperplanes
Associating distinct groups of objects (clusters) with contiguous regions of high probability density (high-density clusters), is a central assumption in statistical and machine learning approaches for the classification of unlabelled data. In unsupervised classification this cluster definition underlies a nonparametric approach known as density clustering. In semi-supervised classification, cl...
متن کاملDeep Transductive Semi-supervised Maximum Margin Clustering
Semi-supervised clustering is an very important topic in machine learning and computer vision. The key challenge of this problem is how to learn a metric, such that the instances sharing the same label are more likely close to each other on the embedded space. However, little attention has been paid to learn better representations when the data lie on non-linear manifold. Fortunately, deep lear...
متن کاملOn a Theory of Nonparametric Pairwise Similarity for Clustering: Connecting Clustering to Classification
Pairwise clustering methods partition the data space into clusters by the pairwise similarity between data points. The success of pairwise clustering largely depends on the pairwise similarity function defined over the data points, where kernel similarity is broadly used. In this paper, we present a novel pairwise clustering framework by bridging the gap between clustering and multi-class class...
متن کاملMaximum Margin Semi-Supervised Learning for Structured Variables
Many real-world classification problems involve the prediction of multiple inter-dependent variables forming some structural dependency. Recent progress in machine learning has mainly focused on supervised classification of such structured variables. In this paper, we investigate structured classification in a semi-supervised setting. We present a discriminative approach that utilizes the intri...
متن کاملComposite Kernel Optimization in Semi-Supervised Metric
Machine-learning solutions to classification, clustering and matching problems critically depend on the adopted metric, which in the past was selected heuristically. In the last decade, it has been demonstrated that an appropriate metric can be learnt from data, resulting in superior performance as compared with traditional metrics. This has recently stimulated a considerable interest in the to...
متن کامل