The model mismatch problem occurs in steganalysis when a binary classifier is trained on objects from one cover source and tested on another: an example of domain adaptation. It is highly realistic because a steganalyst would rarely have access to much or any training data from their opponent, and its consequences can be devastating to classifier accuracy. This paper presents an in-depth study of one particular instance of model mismatch, in a set of images from Flickr using one fixed steganography and steganalysis method, attempting to separate different effects of mismatch in feature space and find methods of mitigation where possible. We also propose new benchmarks for accuracy, which are more appropriate than mean error rates when there are multiple actors and multiple images, and consider the case of 3-valued detectors which also output `don't know'. This pilot study demonstrates that some simple feature-centering and ensemble methods can reduce the mismatch penalty considerably, but not completely remove it.

/pdf/a-mishmash-of-methods-for-mitigating-the-model-mismatch-mess-kwmeeiltvc.pdf

A mishmash of methods for mitigating the model mismatch mess

The Internet is constantly growing in size and becoming more complex. The field of networking is thus continuously progressing to cope with this monumental growth of network traffic. While approaches such as Software Defined Networking (SDN) can provide a centralized control mechanism for network traffic measurement, control, and prediction, still the amount of data received by the SDN controller is huge. To process that data, it has recently been suggested to use Machine Learning (ML). In this paper, we review existing proposal for using ML in an SDN context for traffic measurement (specifically, classification) and traffic prediction. We will especially focus on approaches that use Deep learning (DL) in traffic prediction, which seems to have been mostly untapped by existing surveys. Furthermore, we discuss remaining challenges and suggest future research directions.

Machine Learning and Deep Learning Based Traffic Classification and Prediction in Software Defined Networking

In this paper we propose a novel framework for action recognition based on multiple features for improve action recognition in videos. The fusion of multiple features is important for recognizing actions as often a single feature based representation is not enough to capture the imaging variations (view-point, illumination etc.) and attributes of individuals (size, age, gender etc.). Hence, we use two kinds of features: i) a quantized vocabulary of local spatio-temporal (ST) volumes (cuboids and 2-D SIFT), and ii) the higher-order statistical models of interest points, which aims to capture the global information of the actor. We construct video representation in terms of local space-time features and global features and integrate such representations with hyper-sphere multi-class SVM. Experiments on publicly available datasets show that our proposed approach is effective. An additional experiment shows that using both local and global features provides a richer representation of human action when compared to the use of a single feature type.

Action Recognition by Multiple Features and Hyper-Sphere Multi-class SVM

Discriminative techniques, such as conditional random fields (CRFs) or structure aware maximum-margin techniques (maximum margin Markov networks (M3N), structured output support vector machines (S-SVM)), are state-of-the-art in the prediction of structured data. However, to achieve good results these techniques require complete and reliable ground truth, which is not always available in realistic problems. Furthermore, training either CRFs or margin-based techniques is computationally costly, because the runtime of current training methods depends not only on the size of the training set but also on properties of the output space to which the training samples are assigned.

We propose an alternative model for structured output prediction, Joint Kernel Support Estimation (JKSE), which is rather generative in nature as it relies on estimating the joint probability density of samples and labels in the training set. This makes it tolerant against incomplete or incorrect labels and also opens the possibility of learning in situations where more than one output label can be considered correct.

At the same time, we avoid typical problems of generative models as we do not attempt to learn the full joint probability distribution, but we model only its support in a joint reproducing kernel Hilbert space. As a consequence, JKSE training is possible by an adaption of the classical one-class SVM procedure. The resulting optimization problem is convex and efficiently solvable even with tens of thousands of training examples. A particular advantage of JKSE is that the training speed depends only on the size of the training set, and not on the total size of the label space. No inference step during training is required (as M3N and S-SVM would) nor do we have calculate a partition function (as CRFs do).

Experiments on realistic data show that, for suitable kernel functions, our method works efficiently and robustly in situations that discriminative techniques have problems with or that are computationally infeasible for them.

/pdf/structured-prediction-by-joint-kernel-support-estimation-56a9qseucb.pdf

Structured prediction by joint kernel support estimation

Imbalanced data is a common and serious problem in many biomedical classification tasks. It causes a bias on the training of classifiers and results in lower accuracy of minority classes prediction. This problem has attracted a lot of research interests in the past decade. Unfortunately, most research efforts only concentrate on 2-class problems. In this paper, we study a new method of formulating a multiclass Support Vector Machine (SVM) problem for imbalanced biomedical data to improve the classification performance. The proposed method applies cost-sensitive approach and ramp loss function to the Crammer and Singer multiclass SVM formulation. Experimental results on multiple biomedical datasets show that the proposed solution can effectively cure the problem when the datasets are noisy and highly imbalanced.

Robust Multiclass Classification for Learning from Imbalanced Biomedical Data

Multi-class classification is an important and on-going research subject in machine learning and data mining. In this paper, we propose a new support vector algorithm, called OC-K-SVM, for multi-class classification based on one-class SVM. For k-class problem, this method constructs k classifiers, where each one is trained on data from one class. OC-K-SVM has parameters that enable us to control the number of support vectors and margin errors effectively, which is helpful in improving the accuracy of each classifier. We give some theoretical results concerning the significance of the parameters and show the robustness of classifiers. In addition, we have examined the proposed algorithm on several benchmark data sets, and our preliminary experiments confirm our theoretical conclusions.

/pdf/a-new-multi-class-svm-algorithm-based-on-one-class-svm-22desy9m89.pdf

A New Multi-class SVM Algorithm Based on One-Class SVM

This paper addresses the problem of human action recognition by introducing a new representation of image sequences as a collection of spatio-temporal events that are localized at interest point. The interest points are detected by the SIFT detector and a spatio-temporal interest point detector. We proposed a new bag of words approach to represent videos in two different models. Intersection Kernel Support Vector Machines is used for classification. We also present action classification results on two different datasets. Our results are either comparable to previous published results on these datasets.

Jia Liu

Papers

A New Multi-class SVM Algorithm Based on One-Class SVM

Recognizing human actions using bag-of-features and Intersection Kernel Support Vector Machines