Search results for "Feature learning"
showing 10 items of 16 documents
CostNet: An End-to-End Framework for Goal-Directed Reinforcement Learning
2020
Reinforcement Learning (RL) is a general framework concerned with an agent that seeks to maximize rewards in an environment. The learning typically happens through trial and error using explorative methods, such as \(\epsilon \)-greedy. There are two approaches, model-based and model-free reinforcement learning, that show concrete results in several disciplines. Model-based RL learns a model of the environment for learning the policy while model-free approaches are fully explorative and exploitative without considering the underlying environment dynamics. Model-free RL works conceptually well in simulated environments, and empirical evidence suggests that trial and error lead to a near-opti…
Learning to Navigate in the Gaussian Mixture Surface
2021
In the last years, deep learning models have achieved remarkable generalization capability on computer vision tasks, obtaining excellent results in fine-grained classification problems. Sophisticated approaches based-on discriminative feature learning via patches have been proposed in the literature, boosting the model performances and achieving the state-of-the-art over well-known datasets. Cross-Entropy (CE) loss function is commonly used to enhance the discriminative power of the deep learned features, encouraging the separability between the classes. However, observing the activation map generated by these models in the hidden layer, we realize that many image regions with low discrimin…
Learning Improved Feature Rankings through Decremental Input Pruning for Support Vector Based Drug Activity Prediction
2010
The use of certain machine learning and pattern recognition tools for automated pharmacological drug design has been recently introduced. Different families of learning algorithms and Support Vector Machines in particular have been applied to the task of associating observed chemical properties and pharmacological activities to certain kinds of representations of the candidate compounds. The purpose of this work, is to select an appropriate feature ordering from a large set of molecular descriptors usually used in the domain of Drug Activity Characterization. To this end, a new input pruning method is introduced and assessed with respect to commonly used feature ranking algorithms.
Multimodal Images Classification using Dense SURF, Spectral Information and Support Vector Machine
2019
International audience; The multimodal image classification is a challenging area of image processing which can be used to examine the wall painting in the cultural heritage domain. In such classification, a common space of representation is important. In this paper, we present a new method for multimodal representation learning, by using a pixel-wise feature descriptor named dense Speed Up Robust Features (SURF) combined with the spectral information carried by the pixel. For classification of extracted features we have used support vector machine (SVM). Our database was extracted from acquisition on cultural heritage wall paintings that contain four modalities UV, Visible, IRR and fluores…
Convolutional Neural Networks for the Identification of Regions of Interest in PET Scans: A Study of Representation Learning for Diagnosing Alzheimer…
2017
When diagnosing patients suffering from dementia based on imaging data like PET scans, the identification of suitable predictive regions of interest (ROIs) is of great importance. We present a case study of 3-D Convolutional Neural Networks (CNNs) for the detection of ROIs in this context, just using voxel data, without any knowledge given a priori. Our results on data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) suggest that the predictive performance of the method is on par with that of state-of-the-art methods, with the additional benefit of potential insights into affected brain regions.
A convolutional neural network framework for blind mesh visual quality assessment
2017
In this paper, we propose a new method for blind mesh visual quality assessment using a deep learning approach. To do this, we first extract visual representative features by computing locally curvature and dihedral angles from each distorted mesh. Then, we determine from these features a set of 2D patches which are learned to a convolutional neural network (CNN). The network consists of two convolutional layers with two max-pooling layers. Then, a multilayer perceptron (MLP) with two fully connected layers is integrated to summarize the learned representation into an output node. With this network structure, feature learning and regression are used to predict the quality score of a given d…
Autoencoders and Data Fusion Based Hybrid Health Indicator for Detecting Bearing and Stator Winding Faults in Electric Motors
2018
The main objective of a condition monitoring programs is to track the health status of critical components of a machine. In this paper, a hybrid health indicator is proposed to monitor the health status of bearings and stator winding of a motor. The proposed method is based on a feature learning from deep autoencoders and data fusion. The features can be learned by autoencoders using individual current and vibration signals, and then learning features are fused to make final health indicators. The experimental data from a permanent magnet synchronous motor is used to validate the proposed method. Promising results in detecting faults and severities of the stator and bearing faults at differ…
CNN based Gearbox Fault Diagnosis and Interpretation of Learning Features
2021
Machine learning based fault diagnosis schemes have been intensively proposed to deal with faults diagnosis of rotating machineries such as gearboxes, bearings, and electric motors. However, most of the machine learning algorithms used in fault diagnosis are pattern recognition tools, which can classify given data into two or more classes. The underlined physical phenomena in fault diagnosis are not directly interpretable in machine learning schemes, thus it is usually called black/gray box models. In this study, convolutional neural networks (CNN) machine learning algorithm is proposed to classify gearbox faults, and the learning features of the CNN filters are visualized to understand the…
Combining feature extraction and expansion to improve classification based similarity learning
2017
Abstract Metric learning has been shown to outperform standard classification based similarity learning in a number of different contexts. In this paper, we show that the performance of classification similarity learning strongly depends on the data format used to learn the model. We then present an Enriched Classification Similarity Learning method that follows a hybrid approach that combines both feature extraction and feature expansion. In particular, we propose a data transformation and the use of a set of standard distances to supplement the information provided by the feature vectors of the training samples. The method is compared to state-of-the-art feature extraction and metric lear…
Unsupervised deep feature extraction of hyperspectral images
2014
This paper presents an effective unsupervised sparse feature learning algorithm to train deep convolutional networks on hyperspectral images. Deep convolutional hierarchical representations are learned and then used for pixel classification. Features in lower layers present less abstract representations of data, while higher layers represent more abstract and complex characteristics. We successfully illustrate the performance of the extracted representations in a challenging AVIRIS hyperspectral image classification problem, compared to standard dimensionality reduction methods like principal component analysis (PCA) and its kernel counterpart (kPCA). The proposed method largely outperforms…