This paper provides an introduction to support vector machines (SVMs), kernel Fisher discriminant analysis, and

Abstract. We report on experiments done in an attempt to improve the performance of a music similarity measure which we introduced earlier. The technique aims at comparing music titles on the basis of their global “timbre”, which has many applications in the field of Music Information Retrieval. Such measures of timbre similarity have seen a growing interest lately, and every contribution (including ours) is yet another instantiation of the same basic pattern recognition architecture, only with different algorithm variants and parameters. Most give encouraging results with a little effort, and imply that near-perfect results would just extrapolate by fine-tuning the algorithms ’ parameters. However, such systematic testing over large, interdependent parameter spaces is both difficult and costly, as it requires to work on a whole general meta-database architecture. This paper contributes in two ways to the current state of the art. We report on extensive tests over very many parameters and algorithmic variants, either already envisioned in the literature or not. This leads to an improvement over existing algorithms of about 15 % R-precision. But most importantly, we describe many variants that surprisingly do not lead to any substancial improvement. Moreover, our simulations suggest the existence of a “glass ceiling ” at R-precision about 65 % which cannot probably be overcome by pursuing such variations on the same theme.

Abstract not found

Accurate recognition and tracking of human activities is an important goal of ubiquitous computing. Recent advances in the development of multi-modal wearable sensors enable us to gather rich datasets of human activities. However, the problem of automatically identifying the most useful features for modeling such activities remains largely unsolved. In this paper we present a hybrid approach to recognizing activities, which combines boosting to discriminatively select useful features and learn an ensemble of static classifiers to recognize different activities, with hidden Markov models (HMMs) to capture the temporal regularities and smoothness of activities. We tested the activity recognition system using over 12 hours of wearable-sensor data collected by volunteers in natural unconstrained environments. The models succeeded in identifying a small set of maximally informative features, and were able identify ten different human activities with an accuracy of 95%. 1

Traditional machine learning makes a basic assumption: the training and test data should be under the same distribution. However, in many cases, this identicaldistribution assumption does not hold. The assumption might be violated when a task from one new domain comes, while there are only labeled data from a similar old domain. Labeling the new data can be costly and it would also be a waste to throw away all the old data. In this paper, we present a novel transfer learning framework called TrAdaBoost, which extends boosting-based learning algorithms (Freund & Schapire, 1997). TrAdaBoost allows users to utilize a small amount of newly labeled data to leverage the old data to construct a high-quality classification model for the new data. We show that this method can allow us to learn an accurate model using only a tiny amount of new data and a large amount of old data, even when the new data are not sufficient to train a model alone. We show that TrAdaBoost allows knowledge to be effectively transferred from the old data to the new. The effectiveness of our algorithm is analyzed theoretically and empirically to show that our iterative algorithm can converge well to an accurate model.

Statistical techniques based on hidden Markov Models (HMMs) with Gaussian emission densities have dominated signal processing and pattern recognition literature for the past 20 years. However, HMMs trained using maximum likelihood techniques suffer from an inability to learn discriminative information and are prone to overfitting and over-parameterization. Recent work in machine learning has focused on models, such as the support vector machine (SVM), that automatically control generalization and parameterization as part of the overall optimization process. In this paper, we show that SVMs provide a significant improvement in performance on a static pattern classification task based on the Deterding vowel data. We also describe an application of SVMs to large vocabulary speech recognition, and demonstrate an improvement in error rate on a continuous alphadigit task (OGI Aphadigits) and a large vocabulary conversational speech task (Switchboard). Issues related to the development and optimization of an SVM/HMM hybrid system are discussed.

We present a systematic comparison of machine learning methods applied to the problem of fully automatic recognition of facial expressions. We report results on a series of experiments comparing recognition engines, including AdaBoost, support vector machines, linear discriminant analysis. We also explored feature selection techniques, including the use of AdaBoost for feature selection prior to classification by SVM or LDA. Best results were obtained by selecting a subset of Gabor filters using AdaBoost followed by classification with Support Vector Machines. The system operates in real-time, and obtained 93 % correct generalization to novel subjects for a 7-way forced choice on the Cohn-Kanade expression dataset. The outputs of the classifiers change smoothly as a function of time and thus can be used to measure facial expression dynamics. We applied the system to to fully automated recognition of facial actions (FACS). The present system classifies 17 action units, whether they occur singly or in combination with other actions, with a mean accuracy of 94.8%. We present preliminary results for applying this system to spontaneous facial expressions.

The gap between CPU speeds and the speed of the technologies providing the data is increasing. As a result, latency and bandwidth to needed data is limited by the performance of the storage devices and the networks that connect them to the CPU. Distributed caching techniques are often used to reduce the penalties associated with such caching; however, such techniques need further development to be truly integrated into the network. This paper describes the preliminary design of an adaptive caching scheme using multiple experts, called ACME. ACME is used to manage the replacement policies within distributed caches to further improve the hit rates over static caching techniques. We propose the use of machine learning algorithms to rate and select the current best policies or mixtures of policies via weight updates based on their recent success, allowing each adaptive cache node to tune itself based on the workload it observes. Since no cache databases or synchronization messages are exchanged for adaptivity, the clusters composed of these nodes will be scalable and manageable. We show that static techniques are suboptimal when combined in networks of caches, providing potential for adaptivity to improve performance.

Existing statistical approaches to natural language problems are very coarse approximations to the true complexity of language processing.

We present a systematic comparison of machine learning methods applied to the problem of fully automatic recognition of facial expressions, including AdaBoost, support vector machines, and linear discriminant analysis. Each video-frame is first scanned in real-time to detect approximately upright-frontal faces. The faces found are scaled into image patches of equal size, convolved with a bank of Gabor energy filters, and then passed to a recognition engine that codes facial expressions into 7 dimensions in real time: neutral, anger, disgust, fear, joy, sadness, surprise. We report results on a series of experiments comparing spatial frequency ranges, feature selection techniques, and recognition engines. Best results were obtained by selecting a subset of Gabor filters using AdaBoost and then training Support Vector Machines on the outputs of the filters selected by AdaBoost. The generalization performance to new subjects for a 7-way forced choice was 93% or more correct on two publicly available datasets, the best performance reported so far on these datasets. Surprisingly, registration of internal facial features was not necessary, even though the face detector does not provide precisely registered images. The outputs of the classifier change smoothly as a function of time and thus can be used for unobtrusive motion capture. We developed an end-to-end system that provides facial expression codes at 24 frames per second and animates a computer generated character. In real-time this expression mirror operates down to resolutions of 16 pixels from eye to eye. We also applied the system to fully automated facial action coding.