Showing papers on "Convolutional neural network published in 2002"

Robust face analysis using convolutional neural networks

Abstract: Automatic face analysis has to cope with pose and lighting variations. Especially pose variations are difficult to tackle and many face analysis methods require the use of sophisticated normalization procedures. We propose a data-driven face analysis approach that is not only capable of extracting features relevant to a given face analysis task but is also robust with regard to face location changes and scale variations. This is achieved by deploying convolutional neural networks, which are either trained for facial expression recognition or face identity recognition. Combining the outputs of these networks allows us to obtain a subject dependent or personalized recognition of facial expressions.

A multiple circular path convolution neural network system for detection of mammographic masses

Abstract: A multiple circular path convolution neural network (MCPCNN) architecture specifically designed for the analysis of tumor and tumor-like structures has been constructed. We first divided each suspected tumor area into sectors and computed the defined mass features for each sector independently. These sector features were used on the input layer and were coordinated by convolution kernels of different sizes that propagated signals to the second layer in the neural network system. The convolution kernels were trained, as required, by presenting the training cases to the neural network. In this study, randomly selected mammograms were processed by a dual morphological enhancement technique. Radiodense areas were isolated and were delineated using a region growing algorithm. The boundary of each region of interest was then divided into 36 sectors using 36 equi-angular dividers radiated from the center of the region. A total of 144 Breast Imaging-Reporting and Data System-based features (i.e., four features per sector for 36 sectors) were computed as input values for the evaluation of this newly invented neural network system. The overall performance was 0.78-0.80 for the areas (A/sub z/) under the receiver operating characteristic curves using the conventional feed-forward neural network in the detection of mammographic masses. The performance was markedly improved with A/sub z/ values ranging from 0.84 to 0.89 using the MCPCNN. This paper does not intend to claim the best mass detection system. Instead it reports a potentially better neural network structure for analyzing a set of the mass features defined by an investigator.

A neural architecture for fast and robust face detection

Abstract: In this paper, we present a connectionist approach for detecting and precisely localizing semi-frontal human faces in complex images, making no assumption about the content or the lighting conditions of the scene, or about the size or the appearance of the faces. We propose a convolutional neural network architecture designed to recognize strongly variable face patterns directly from pixel images with no preprocessing, by automatically synthesizing its own set of feature extractors from a large training set of faces. We present in details the optimized design of our architecture, our learning strategy and the resulting process of face detection. We also provide experimental results to demonstrate the robustness of our approach and its capability to precisely detect extremely variable faces in uncontrolled environments.

Head-pose invariant facial expression recognition using convolutional neural networks

Abstract: Automatic face analysis has to cope with pose and lighting variations. Especially pose variations are difficult to tackle and many face analysis methods require the use of sophisticated normalization and initialization procedures. We propose a data-driven face analysis approach that is not only capable of extracting features relevant to a given face analysis task, but is also more robust with regard to face location changes and scale variations when compared to classical methods such as e.g. MLPs. Our approach is based on convolutional neural networks that use multi-scale feature extractors, which allow for improved facial expression recognition results with faces subject to in-plane pose variations.

Multiscale Facial Expression Recognition Using Convolutional Neural Networks.

Abstract: Automatic face analysis has to cope with pose and lighting variations. Especially pose variations are difficult to tackle and many face analysis methods require the use of sophisticated normalization procedures. We propose a datadriven face analysis approach that is not only capable of extracting features relevant to a given face analysis task, but is also robust with regard to face location changes and scale variations. This is achieved by deploying convolutional neural networks. We show that the use of multiscale feature extractors and whole-field feature map summing neurons allow to improve facial expression recognition results, especially with test sets that feature scale, respectively, translation changes.

Multi-modular architecture based on convolutional neural networks for online handwritten character recognition

Abstract: In this paper, several convolutional neural network architectures are investigated for online isolated handwritten character recognition (Latin alphabet). Two main architectures have been developed and optimised. The first one, a TDNN, processes online features extracted from the character. The second one, a SDNN, relies on the off-line bitmaps reconstructed from the trajectory of the pen. Moreover, an hybrid architecture called SDTDNN has been derived, it allows the combination of on-line and off-line recognisers. Such a combination seems to be very promising to enhance the character recognition rate. This type of shared weights neural networks introduces the notion of receptive field, local extraction and it allows to restrain the number of free parameters in opposition to classic techniques such as multi-layer perceptron. Results on UNIPEN and IRONOFF databases for online recognition are reported, while the MNIST database has been used for the off-line classifier.

A parallel neural processor for real-time applications

Abstract: Training and testing artificial neural networks can be challenging and time-consuming. Experiments with two real-time applications were performed to compare three approaches for implementing a multilayer perceptron neural network. In both applications, the special-purpose processor performed best.

A hybrid model for invariant and perceptual texture mapping

Abstract: Texture is an important visual feature for computer vision tasks. In applications such as image retrieval and computer image understanding, texture similarity should be measured in a manner that is invariant to texture scale and orientation, as well as consistent with human perception. However, most existing computational features and similarity measures are not perceptually consistent. A solution is to map textures into an invariant and perceptual space such that similarity measured in the space is perceptually consistent. The paper presents a hybrid method, using a convolutional neural network and SVM, to perform the invariant and perceptual mapping. Test results show that it's overall performance is better than that of an individual neural network. and a SVM.

Facial expression analysis using shape and motion information extracted by convolutional neural networks

Abstract: We discuss a neural networks-based face analysis approach that is able to cope with faces subject to pose and lighting variations. Especially head pose variations are difficult to tackle and many face analysis methods require the use of sophisticated normalization procedures. Data-driven shape and motion-based face analysis approaches are introduced that are not only capable of extracting features relevant to a given face analysis task, but are also robust with regard to translation and scale variations. This is achieved by deploying convolutional and time-delayed neural networks, which are either trained for face shape deformation or facial motion analysis.

Convolutional Neural Networks for Radar Detection

Abstract: The use of convolutional neural networks (CNN's) for radar detection is evaluated. The detector includes a time-frequency block that has been implemented by the Wigner-Ville distribution and the Short-Time Fourier Transform to test the suitability of both techniques. The CNN detectors are compared with the classic multilayer perceptron and with several traditional non-neural detectors. Preliminary results are shown using non-correlated and correlated Rayleigh-envelope clutter.

Robust Face Detection Based on Convolutional Neural Networks

Abstract: Automatic face detection in digital video is becoming a very important research topic, due to its wide range of applications, such as security access control, model-based video coding or content-based video indexing. In this paper, we present a connectionist approach for detecting and precisely localizing semi-frontal human faces in complex images, making no assumption on the content or the lighting conditions of the scene, neither on the size, the orientation, and the appearance of the faces. Unlike other systems depending on a hand-crafted feature detection stage, followed by a feature classification stage, we propose a convolutional neural network architecture designed to recognize strongly variable face patterns directly from pixel images with no preprocessing, by automatically synthesizing its own set of feature extractors from a large training set of faces. Moreover, the use of receptive fields, shared weights and spatial subsampling in such a neural model provides some degrees of invariance to translation, rotation, scale, and deformation of the face patterns. We present in details the optimized design of our architecture and our learning strategy. Then, we present the process of face detection using this architecture. Finally, we provide experimental results to demonstrate the robustness of our approach and its capability to precisely detect extremely variable faces in uncontrolled environment.

Mutliscale Facial Expression Recognition using Convolutional Neural Networks

Abstract: Automatic face analysis has to cope with pose and lighting variations. Especially pose variations are difficult to tackle and many face analysis methods require the use of sophisticated normalization procedures. We propose a data-driven face analysis approach that is not only capable of extracting features relevant to a given face analysis task, but is also robust with regard to face location changes and scale variations. This is achieved by deploying convolutional neural networks. We show that the use of multi-scale feature extractors and whole-field feature map summing neurons allow to improve facial expression recognition results, especially with test sets that feature scale, respectively, translation changes.

Convolutional neural networks for radar detection

Abstract: The use of convolutional neural networks (CNN's) for radar detection is evaluated The detector includes a time-frequency block that has been implemented by the Wigner-Ville distribution and the Short-Time Fourier Transform to test the suitability of both techniques The CNN detectors are compared with the classic multilayer perceptron and with several traditional non-neural detectors Preliminary results are shown using non-correlated and correlated Rayleigh-envelope clutter

A novel artificial neural network trained using evolutionary algorithms for reinforcement learning

Abstract: This paper discusses the development of a novel pattern recognition system using artificial neural networks (ANNs) and evolutionary algorithms for reinforcement learning (EARL). The network is based on neuronal interactions involved in identification of prey and predator in toads. The distributed neural network (DNN) is capable of recognizing and classifying various features. The lateral inhibition between the output neurons helps the network in the classification process - similar to the gate in gating network. The results obtained are compared with standard neural network architectures and learning algorithms.

Pattern learning by multilayer neural networks trained by a moderatism-based new algorithm

Abstract: There are many learning algorithms for artificial neural networks today. However, most of these algorithms do not consider the learning characteristics of living creatures. We propose a new learning algorithm that is based on such a learning characteristic called "Moderatism". This new rule shows superiority over the well-known error backpropagation in some pattern learning experiments. Also the inclusion of the error of Moderatism in error backpropagation brings better learning performance.

Self-organizing neural networks using adaptive neurons

Abstract: In this paper, we propose a new kind of neural network having modular structure, neural network with adaptive neurons. Each module is equivalent to an adaptive neuron, which consists of a multi-layer neural network with sigmoid neurons. We develop an algorithm by which the network can automatically adjust its complexity according to the given problem. The proposed network is compared with the project pursuit learning network (PPLN), which is a popular modular architecture. The experimental results demonstrate that the proposed network architecture outperforms the PPLN on four regression problems.