Topic
Multiple kernel learning
About: Multiple kernel learning is a research topic. Over the lifetime, 1630 publications have been published within this topic receiving 56082 citations.
Papers published on a yearly basis
Papers
More filters
TL;DR: This paper integrates the localizing gating model into the previous work named MultiK-MHKS that is an effective multiple empirical kernel learning named MLEKL and establishes a localized formulation in the empiricalkernel learning framework.
Abstract: The traditional multiple kernel learning (MKL) is usually based on implicit kernel mapping and adopts a certain combination of kernels instead of a single kernel. MKL has been demonstrated to have a significant advantage to the single-kernel learning. Although MKL sets different weights to different kernels, the weights are not changed over the whole input space. This weight setting might not been fit for those data with some underlying local distributions. In order to solve this problem, Gonen and Alpaydin (2008) introduced a localizing gating model into the traditional MKL framework so as to assign different weights to a kernel in different regions of the input space. In this paper, we also integrate the localizing gating model into our previous work named MultiK-MHKS that is an effective multiple empirical kernel learning. In doing so, we can get multiple localized empirical kernel learning named MLEKL. Our contribution is that we first establish a localized formulation in the empirical kernel learning framework. The experimental results on benchmark data sets validate the effectiveness of the proposed MLEKL.
6 citations
TL;DR: From experimental results on some data sets, it is shown that the proposed MKL framework based on the notion of gaussianity followed by FDA offers strong classification power.
Abstract: Kernel methods are known to be effective for nonlinear multivariate analysis. One of the main issues in the practical use of kernel methods is the selection of kernel. There have been a lot of studies on kernel selection and kernel learning. Multiple kernel learning (MKL) is one of the promising kernel optimization approaches. Kernel methods are applied to various classifiers including Fisher discriminant analysis (FDA). FDA gives the Bayes optimal classification axis if the data distribution of each class in the feature space is a gaussian with a shared covariance structure. Based on this fact, an MKL framework based on the notion of gaussianity is proposed. As a concrete implementation, an empirical characteristic function is adopted to measure gaussianity in the feature space associated with a convex combination of kernel functions, and two MKL algorithms are derived. From experimental results on some data sets, we show that the proposed kernel learning followed by FDA offers strong classification power.
6 citations
22 Aug 2016
TL;DR: This paper proposes a new sentiment analysis method based on Simple Multiple Kernel Learning (SimpleMKL), which not only outperforms some common methods, like SVM, Naive Bayes, KNN, Random Forest, Adaboost, etc., using the image features of colour, hog, texture, but also outper performs some state-of-the-art methods.
Abstract: Image is one of the most important means to express users' emotions on microblogging, like Sina Weibo. More and more people post only images on it, due to the fast and convenient nature of image. Taking a post only using images on microblogging has been a new tendency. Most existing studies about sentiment analysis on microblogging focus on the text, or integrate image as an auxiliary information into text, so they are not applicable in this scenario. Although a few methods related to sentiment analysis for image have been proposed, most of them either ignore the semantic gap between low-level visual features and higher-level image sentiments, or require a lot of textual information in the phases of both training and inference. This paper proposes a new sentiment analysis method based on Simple Multiple Kernel Learning (SimpleMKL). Specifically, textual information as a sort of sufficiently emotional source data, we can use it to promote the ability via SimpleMKL to classify images. And once we get the image classifier, none of texts are needed when predicting other unlabelled images. Experimental results show that our proposed method can improve the performance significantly on data we crawled and labelled from Sina Weibo. We find that our method not only outperforms some common methods, like SVM, Naive Bayes, KNN, Random Forest, Adaboost, etc., using the image features of colour, hog, texture, but also outperforms some state-of-the-art methods.
6 citations
03 Dec 2015
TL;DR: Experimental results show that the extracted trajectories can describe the movement process of object and the proposed method with multiple kernel learning obtains good performance.
Abstract: Trajectories extracted by previous methods for human action recognition contain irrelevant changes, and the Orientation-Magnitude descriptors of their shapes lack the robustness to camera motion. To solve these problems, action recognition by tracking salient relative motion points is proposed in this paper. Firstly, motion boundary detector which suppresses the camera constant motion is utilized to extract motion features. After processing the detected boundaries by the adaptive threshold, the super-pixels that contain salient points are defined as relative motion regions. Then tracking the points within super-pixels is to generate trajectories. For the trajectory shape, the pre-defined orientation assignments with coarse-to-fine quantization levels are used to produce orientation statistics. Finally, the descriptors of oriented gradient, motion boundary, oriented statistic and their combination are adopted to represent action videos, respectively. On the benchmark KTH and UCF-sports action datasets, experimental results show that the extracted trajectories can describe the movement process of object. Compared with the conventional algorithms, our method with multiple kernel learning obtains good performance.
6 citations
25 Jun 2011
TL;DR: Experimental results show that MK-SVMs achieve promising performance on the heavily skewed dataset by means of a re-balanced strategy, and the extracted rules achieves high coverage and low false-alarm with small number of preconditions.
Abstract: The in-depth understanding of customers' behavior is a crucial means in CRM for identifying its driving force and developing effective and personalized service activities. Relatively little research notices that it is key important in the competitive market to build comprehensible customer churn prediction models which can provide enterprises explicit customer behavior patterns. In this paper, a multiple kernel support vector machines (MK-SVMs) based customer churn prediction model is proposed to encapsulate three knowledge discovery tasks, which are feature selection, class prediction and decision rule extraction, into a whole framework A two-stage iteration of two convex optimization problems is designed for simultaneously feature selection and class prediction. Based on the selected features, support vectors are used to extract decision rules. An open CRM dataset is used to evaluate the performance of this approach. Experimental results show that MK-SVMs achieve promising performance on the heavily skewed dataset by means of a re-balanced strategy, and the extracted rules achieves high coverage and low false-alarm with small number of preconditions.
6 citations