PyTorchKR Papers

최신 논문

총 12,350편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1307.1391

Quiet in Class: Classification, Noise and the Dendritic Cell Algorithm

Theoretical analyses of the Dendritic Cell Algorithm (DCA) have yielded several criticisms about its underlying structure and operation. As a result, several alterations and fixes have been suggested in the literature to correct for these findings. A contribution of this work is to investigate the effects of replacing the classification stage of the DCA (which is known to be flawed) with a traditional machine learning technique. This work goes on to question the merits of those unique properties of the DCA that are yet to be thoroughly analysed. If none of these properties can be found to have a benefit over traditional approaches, then "fixing" the DCA is arguably less efficient than simply creating a new algorithm. This work examines the dynamic filtering property of the DCA and questions the utility of this unique feature for the anomaly detection problem. It is found that this feature, while advantageous for noisy, time-ordered classification, is not as useful as a traditional static filter for processing a synthetic dataset. It is concluded that there are still unique features of the DCA left to investigate. Areas that may be of benefit to the Artificial Immune Systems community are suggested.

arXiv:1307.1387

Examining the Classification Accuracy of TSVMs with ?Feature Selection in Comparison with the GLAD Algorithm

Gene expression data sets are used to classify and predict patient diagnostic categories. As we know, it is extremely difficult and expensive to obtain gene expression labelled examples. Moreover, conventional supervised approaches cannot function properly when labelled data (training examples) are insufficient using Support Vector Machines (SVM) algorithms. Therefore, in this paper, we suggest Transductive Support Vector Machines (TSVMs) as semi-supervised learning algorithms, learning with both labelled samples data and unlabelled samples to perform the classification of microarray data. To prune the superfluous genes and samples we used a feature selection method called Recursive Feature Elimination (RFE), which is supposed to enhance the output of classification and avoid the local optimization problem. We examined the classification prediction accuracy of the TSVM-RFE algorithm in comparison with the Genetic Learning Across Datasets (GLAD) algorithm, as both are semi-supervised learning methods. Comparing these two methods, we found that the TSVM-RFE surpassed both a SVM using RFE and GLAD.

arXiv:1307.0781

Distributed Online Big Data Classification Using Context Information

Distributed, online data mining systems have emerged as a result of applications requiring analysis of large amounts of correlated and high-dimensional data produced by multiple distributed data sources. We propose a distributed online data classification framework where data is gathered by distributed data sources and processed by a heterogeneous set of distributed learners which learn online, at run-time, how to classify the different data streams either by using their locally available classification functions or by helping each other by classifying each other's data. Importantly, since the data is gathered at different locations, sending the data to another learner to process incurs additional costs such as delays, and hence this will be only beneficial if the benefits obtained from a better classification will exceed the costs. We model the problem of joint classification by the distributed and heterogeneous learners from multiple data sources as a distributed contextual bandit problem where each data is characterized by a specific context. We develop a distributed online learning algorithm for which we can prove sublinear regret. Compared to prior work in distributed online data mining, our work is the first to provide analytic regret results characterizing the performance of the proposed algorithm.

arXiv:1306.6802

Evaluation Measures for Hierarchical Classification: a unified view and novel approaches

Hierarchical classification addresses the problem of classifying items into a hierarchy of classes. An important issue in hierarchical classification is the evaluation of different classification algorithms, which is complicated by the hierarchical relations among the classes. Several evaluation measures have been proposed for hierarchical classification using the hierarchy in different ways. This paper studies the problem of evaluation in hierarchical classification by analyzing and abstracting the key components of the existing performance measures. It also proposes two alternative generic views of hierarchical evaluation and introduces two corresponding novel measures. The proposed measures, along with the state-of-the art ones, are empirically tested on three large datasets from the domain of text classification. The empirical results illustrate the undesirable behavior of existing approaches and how the proposed methods overcome most of these methods across a range of cases.

arXiv:1302.1043

The price of bandit information in multiclass online classification

We consider two scenarios of multiclass online learning of a hypothesis class $H\subseteq Y^X$. In the {\em full information} scenario, the learner is exposed to instances together with their labels. In the {\em bandit} scenario, the true label is not exposed, but rather an indication whether the learner's prediction is correct or not. We show that the ratio between the error rates in the two scenarios is at most $8\cdot|Y|\cdot \log(|Y|)$ in the realizable case, and $\tilde{O}(\sqrt{|Y|})$ in the agnostic case. The results are tight up to a logarithmic factor and essentially answer an open question from (Daniely et. al. - Multiclass learnability and the erm principle). We apply these results to the class of $\gamma$-margin multiclass linear classifiers in $\reals^d$. We show that the bandit error rate of this class is $\tilde{\Theta}(\frac{|Y|}{\gamma^2})$ in the realizable case and $\tilde{\Theta}(\frac{1}{\gamma}\sqrt{|Y|T})$ in the agnostic case. This resolves an open question from (Kakade et. al. - Efficient bandit algorithms for online multiclass prediction).

arXiv:1307.5730

A New Strategy of Cost-Free Learning in the Class Imbalance Problem

In this work, we define cost-free learning (CFL) formally in comparison with cost-sensitive learning (CSL). The main difference between them is that a CFL approach seeks optimal classification results without requiring any cost information, even in the class imbalance problem. In fact, several CFL approaches exist in the related studies, such as sampling and some criteria-based pproaches. However, to our best knowledge, none of the existing CFL and CSL approaches are able to process the abstaining classifications properly when no information is given about errors and rejects. Based on information theory, we propose a novel CFL which seeks to maximize normalized mutual information of the targets and the decision outputs of classifiers. Using the strategy, we can deal with binary/multi-class classifications with/without abstaining. Significant features are observed from the new strategy. While the degree of class imbalance is changing, the proposed strategy is able to balance the errors and rejects accordingly and automatically. Another advantage of the strategy is its ability of deriving optimal rejection thresholds for abstaining classifications and the "equivalent" costs in binary classifications. The connection between rejection thresholds and ROC curve is explored. Empirical investigation is made on several benchmark data sets in comparison with other existing approaches. The classification results demonstrate a promising perspective of the strategy in machine learning.

arXiv:1307.5702

Is Bottom-Up Attention Useful for Scene Recognition?

The human visual system employs a selective attention mechanism to understand the visual world in an eficient manner. In this paper, we show how computational models of this mechanism can be exploited for the computer vision application of scene recognition. First, we consider saliency weighting and saliency pruning, and provide a comparison of the performance of different attention models in these approaches in terms of classification accuracy. Pruning can achieve a high degree of computational savings without significantly sacrificing classification accuracy. In saliency weighting, however, we found that classification performance does not improve. In addition, we present a new method to incorporate salient and non-salient regions for improved classification accuracy. We treat the salient and non-salient regions separately and combine them using Multiple Kernel Learning. We evaluate our approach using the UIUC sports dataset and find that with a small training size, our method improves upon the classification accuracy of the baseline bag of features approach.

arXiv:1307.5599

Performance comparison of State-of-the-art Missing Value Imputation Algorithms on Some Bench mark Datasets

Decision making from data involves identifying a set of attributes that contribute to effective decision making through computational intelligence. The presence of missing values greatly influences the selection of right set of attributes and this renders degradation in classification accuracies of the classifiers. As missing values are quite common in data collection phase during field experiments or clinical trails appropriate handling would improve the classifier performance. In this paper we present a review of recently developed missing value imputation algorithms and compare their performance on some bench mark datasets.

arXiv:1307.5497

A scalable stage-wise approach to large-margin multi-class loss based boosting

We present a scalable and effective classification model to train multi-class boosting for multi-class classification problems. Shen and Hao introduced a direct formulation of multi- class boosting in the sense that it directly maximizes the multi- class margin [C. Shen and Z. Hao, "A direct formulation for totally-corrective multi- class boosting", in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2011]. The major problem of their approach is its high computational complexity for training, which hampers its application on real-world problems. In this work, we propose a scalable and simple stage-wise multi-class boosting method, which also directly maximizes the multi-class margin. Our approach of- fers a few advantages: 1) it is simple and computationally efficient to train. The approach can speed up the training time by more than two orders of magnitude without sacrificing the classification accuracy. 2) Like traditional AdaBoost, it is less sensitive to the choice of parameters and empirically demonstrates excellent generalization performance. Experimental results on challenging multi-class machine learning and vision tasks demonstrate that the proposed approach substantially improves the convergence rate and accuracy of the final visual detector at no additional computational cost compared to existing multi-class boosting.

arXiv:1307.4717

Content Based Image Retrieval System using Feature Classification with Modified KNN Algorithm

Feature means countenance, remote sensing scene objects with similar characteristics, associated to interesting scene elements in the image formation process. They are classified into three types in image processing, that is low, middle and high. Low level features are color, texture and middle level feature is shape and high level feature is semantic gap of objects. An image retrieval system is a computer system for browsing, searching and retrieving images from a large image database. Content Based Image Retrieval is a technique which uses visual features of image such as color, shape, texture to search user required image from large image database according to user requests in the form of a query. MKNN is an enhancing method of KNN. The proposed KNN classification is called MKNN. MKNN contains two parts for processing, they are validity of the train samples and applying weighted KNN. The validity of each point is computed according to its neighbors. In our proposal, Modified K-Nearest Neighbor can be considered a kind of weighted KNN so that the query label is approximated by weighting the neighbors of the query.

arXiv:1307.3626

Learning an Integrated Distance Metric for Comparing Structure of Complex Networks

Graph comparison plays a major role in many network applications. We often need a similarity metric for comparing networks according to their structural properties. Various network features - such as degree distribution and clustering coefficient - provide measurements for comparing networks from different points of view, but a global and integrated distance metric is still missing. In this paper, we employ distance metric learning algorithms in order to construct an integrated distance metric for comparing structural properties of complex networks. According to natural witnesses of network similarities (such as network categories) the distance metric is learned by the means of a dataset of some labeled real networks. For evaluating our proposed method which is called NetDistance, we applied it as the distance metric in K-nearest-neighbors classification. Empirical results show that NetDistance outperforms previous methods, at least 20 percent, with respect to precision.

arXiv:1112.5246

Combining One-Class Classifiers via Meta-Learning

Selecting the best classifier among the available ones is a difficult task, especially when only instances of one class exist. In this work we examine the notion of combining one-class classifiers as an alternative for selecting the best classifier. In particular, we propose two new one-class classification performance measures to weigh classifiers and show that a simple ensemble that implements these measures can outperform the most popular one-class ensembles. Furthermore, we propose a new one-class ensemble scheme, TUPSO, which uses meta-learning to combine one-class classifiers. Our experiments demonstrate the superiority of TUPSO over all other tested ensembles and show that the TUPSO performance is statistically indistinguishable from that of the hypothetical best classifier.

arXiv:1308.0658

Exploring The Contribution of Unlabeled Data in Financial Sentiment Analysis

With the proliferation of its applications in various industries, sentiment analysis by using publicly available web data has become an active research area in text classification during these years. It is argued by researchers that semi-supervised learning is an effective approach to this problem since it is capable to mitigate the manual labeling effort which is usually expensive and time-consuming. However, there was a long-term debate on the effectiveness of unlabeled data in text classification. This was partially caused by the fact that many assumptions in theoretic analysis often do not hold in practice. We argue that this problem may be further understood by adding an additional dimension in the experiment. This allows us to address this problem in the perspective of bias and variance in a broader view. We show that the well-known performance degradation issue caused by unlabeled data can be reproduced as a subset of the whole scenario. We argue that if the bias-variance trade-off is to be better balanced by a more effective feature selection method unlabeled data is very likely to boost the classification performance. We then propose a feature selection framework in which labeled and unlabeled training samples are both considered. We discuss its potential in achieving such a balance. Besides, the application in financial sentiment analysis is chosen because it not only exemplifies an important application, the data possesses better illustrative power as well. The implications of this study in text classification and financial sentiment analysis are both discussed.

arXiv:1307.8430

Fast Simultaneous Training of Generalized Linear Models (FaSTGLZ)

We present an efficient algorithm for simultaneously training sparse generalized linear models across many related problems, which may arise from bootstrapping, cross-validation and nonparametric permutation testing. Our approach leverages the redundancies across problems to obtain significant computational improvements relative to solving the problems sequentially by a conventional algorithm. We demonstrate our fast simultaneous training of generalized linear models (FaSTGLZ) algorithm on a number of real-world datasets, and we run otherwise computationally intensive bootstrapping and permutation test analyses that are typically necessary for obtaining statistically rigorous classification results and meaningful interpretation. Code is freely available at http://liinc.bme.columbia.edu/fastglz.

arXiv:1307.8012

A Study on Classification in Imbalanced and Partially-Labelled Data Streams

The domain of radio astronomy is currently facing significant computational challenges, foremost amongst which are those posed by the development of the world's largest radio telescope, the Square Kilometre Array (SKA). Preliminary specifications for this instrument suggest that the final design will incorporate between 2000 and 3000 individual 15 metre receiving dishes, which together can be expected to produce a data rate of many TB/s. Given such a high data rate, it becomes crucial to consider how this information will be processed and stored to maximise its scientific utility. In this paper, we consider one possible data processing scenario for the SKA, for the purposes of an all-sky pulsar survey. In particular we treat the selection of promising signals from the SKA processing pipeline as a data stream classification problem. We consider the feasibility of classifying signals that arrive via an unlabelled and heavily class imbalanced data stream, using currently available algorithms and frameworks. Our results indicate that existing stream learners exhibit unacceptably low recall on real astronomical data when used in standard configuration; however, good false positive performance and comparable accuracy to static learners, suggests they have definite potential as an on-line solution to this particular big data challenge.

arXiv:1307.7429

Participation anticipating in elections using data mining methods

Anticipating the political behavior of people will be considerable help for election candidates to assess the possibility of their success and to be acknowledged about the public motivations to select them. In this paper, we provide a general schematic of the architecture of participation anticipating system in presidential election by using KNN, Classification Tree and Na\"ive Bayes and tools orange based on crisp which had hopeful output. To test and assess the proposed model, we begin to use the case study by selecting 100 qualified persons who attend in 11th presidential election of Islamic republic of Iran and anticipate their participation in Kohkiloye & Boyerahmad. We indicate that KNN can perform anticipation and classification processes with high accuracy in compared with two other algorithms to anticipate participation.

arXiv:1307.6937

A Novel Architecture For Question Classification Based Indexing Scheme For Efficient Question Answering

Question answering system can be seen as the next step in information retrieval, allowing users to pose question in natural language and receive compact answers. For the Question answering system to be successful, research has shown that the correct classification of question with respect to the expected answer type is requisite. We propose a novel architecture for question classification and searching in the index, maintained on the basis of expected answer types, for efficient question answering. The system uses the criteria for Answer Relevance Score for finding the relevance of each answer returned by the system. On analysis of the proposed system, it has been found that the system has shown promising results than the existing systems based on question classification.

arXiv:1307.6522

When is the majority-vote classifier beneficial?

In his seminal work, Schapire (1990) proved that weak classifiers could be improved to achieve arbitrarily high accuracy, but he never implied that a simple majority-vote mechanism could always do the trick. By comparing the asymptotic misclassification error of the majority-vote classifier with the average individual error, we discover an interesting phase-transition phenomenon. For binary classification with equal prior probabilities, our result implies that, for the majority-vote mechanism to work, the collection of weak classifiers must meet the minimum requirement of having an average true positive rate of at least 50% and an average false positive rate of at most 50%.

arXiv:1305.3814

Multi-View Learning for Web Spam Detection

Spam pages are designed to maliciously appear among the top search results by excessive usage of popular terms. Therefore, spam pages should be removed using an effective and efficient spam detection system. Previous methods for web spam classification used several features from various information sources (page contents, web graph, access logs, etc.) to detect web spam. In this paper, we follow page-level classification approach to build fast and scalable spam filters. We show that each web page can be classified with satisfiable accuracy using only its own HTML content. In order to design a multi-view classification system, we used state-of-the-art spam classification methods with distinct feature sets (views) as the base classifiers. Then, a fusion model is learned to combine the output of the base classifiers and make final prediction. Results show that multi-view learning significantly improves the classification performance, namely AUC by 22%, while providing linear speedup for parallel execution.

arXiv:1202.2564

A better Beta for the H measure of classification performance

The area under the ROC curve is widely used as a measure of performance of classification rules. However, it has recently been shown that the measure is fundamentally incoherent, in the sense that it treats the relative severities of misclassifications differently when different classifiers are used. To overcome this, Hand (2009) proposed the $H$ measure, which allows a given researcher to fix the distribution of relative severities to a classifier-independent setting on a given problem. This note extends the discussion, and proposes a modified standard distribution for the $H$ measure, which better matches the requirements of researchers, in particular those faced with heavily unbalanced datasets, the $Beta(\pi_1+1,\pi_0+1)$ distribution. [Preprint submitted at Pattern Recognition Letters]