The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
Image steganalysis is to discriminate innocent images and those suspected images with hidden messages. In this paper, we propose a unified Convolutional Neural Network (CNN) model for this task. In order to reliably detect modern steganographic algorithms, we design the proposed model from two aspects. For the first, different from existing CNN based steganalytic algorithms that use a predefined highpass...
State-of-the-art convolutional neural networks are enormously costly in both compute and memory, demanding massively parallel GPUs for execution. Such networks strain the computational capabilities and energy available to embedded and mobile processing platforms, restricting their use in many important applications. In this paper, we propose BCNN with Separable Filters (BCNNw/SF), which applies Singular...
According to some estimates of World Health Organization (WHO), in 2014, more than 1.9 billion adults aged 18 years and older were overweight. Overall, about 13% of the world's adult population (11% of men and 15% of women) were obese. 39% of adults aged 18 years and over (38% of men and 40% of women) were overweight. The worldwide prevalence of obesity more than doubled between 1980 and 2014. The...
Convolutional Neural Networks have established a new standard in many machine learning applications not only in image but also in audio processing. In this contribution we investigate the interplay between the primary representation mapping a raw audio signal to some kind of image (feature) and the convolutional layers of an ensuing neural network. We introduce a new notion of equivalence of feature-network...
The aim of scene semantic segmentation is to label each pixel with a class which it belongs to in high level cognition. State-of-art works mainly adapt convolutional neural networks originally designed for image classification to make dense prediction. However the inner structure of scene itself and its stuff is more flexible and variable, which is distinct from the objects in image classification...
We present a novel deep learning framework for crowd counting by learning a perspective-embedded deconvolution network. Perspective is an inherent property of most surveillance scenes. Unlike the traditional approaches that exploit the perspective as a separate normalization, we propose to fuse the perspective into a deconvolution network, aiming to obtain a robust, accurate and consistent crowd density...
This paper proposes a novel shape feature extractor named Contour-SIFT along with a matching method that computes the similarity between two set of proposed descriptors. It allows a shape to be recognized based on automatically located outstanding local features on its contour, which are extracted from 1-D signal representations of different smoothing scales. The algorithm describes each local feature...
In this paper, we focus on constructing an accurate super resolution system based on multiple Convolution Neural Networks (CNNs). Each individual CNN is trained separately with different network structure. A Context-wise Network Fusion (CNF) approach is proposed to integrate the outputs of individual networks by additional convolution layers. With fine-tuning the whole fused network, the accuracy...
In this work, we propose CLass-Enhanced Attentive Response (CLEAR): an approach to visualize and understand the decisions made by deep neural networks (DNNs) given a specific input. CLEAR facilitates the visualization of attentive regions and levels of interest of DNNs during the decision-making process. It also enables the visualization of the most dominant classes associated with these attentive...
Convolutional Neural Network (CNN) has become one of the most successful technologies for visual classification and other applications. As CNN models continue to evolve and adopt different kernel sizes in various applications, it is necessary for the hardware architecture to support reconfigurability. Previous FPGAs and programmable ASICs are fine-grained reconfigurable but with energy efficiency...
This paper presents an approach to enhance the performance of machine learning applications based on hardware acceleration. This approach is based on parameterised architectures designed for Convolutional Neural Network (CNN) and Support Vector Machine (SVM), and the associated design flow common to both. This approach is illustrated by two case studies including object detection and satellite data...
AICNN architecture is presented in this work to map the state-of-the-art machine-learning algorithms of CNN to power-constrained embedded hardware. As the combination of analog-to-information conversion and typical CNN algorithms, AICNN can realize ultra-highly efficient computation by using massive parallel analog signal processing circuits, which could also significantly reduce ADC devices cost...
Convolutional neural networks (CNNs) have emerged as one of the most successful machine learning technologies for image and video processing. The most computationally-intensive parts of CNNs are the convolutional layers, which convolve multi-channel images with multiple kernels. A common approach to implementing convolutional layers is to expand the image into a column matrix (im2col) and perform...
State-of-the-art CNN models for Image recognition use deep networks with small filters instead of shallow networks with large filters, because the former requires fewer weights. In the light of above trend, we present a fast and efficient FPGA based convolution engine to accelerate CNN models over small filters. The convolution engine implements Winograd minimal filtering algorithm to reduce the number...
In this work, we conduct research on optimizing schemes for the RRAM-based implementation of CNN. Our main achievements contain: 1) A concrete CNN circuit and corresponding operation methods are developed. 2) Quantification methods for utilizing binary or multilevel RRAM as synapses are proposed, and our CNN performs with 98% accuracy on the MNIST dataset using multilevel RRAM and 97% accuracy using...
In this paper, we propose a hardware computing architecture for face detection that classifies an image as a face or non-face. The computing architecture is first designed, modeled and tested in MATLAB Simulink using Xilinx block set and was later tested using a Virtex-6 FPGA ML605 Evaluation Kit. The system uses learned filters which were previously extracted by training on a set of face and non-face...
In recommender systems, numerous efforts have been made on utilizing textual information in matrix factorization to alleviate the problem of data sparsity. Recently, some of the works have explored neural networks to go for an in-depth understanding of textual item content, and further generate more accurate item latent models. These works achieve impressive effectiveness on performing recommendations...
In order to defend adversarial attacks in computer vision models, the conventional approach arises on actively incorporate such samples into the training datasets. Nonetheless, the manual production of adversarial samples is painful and labor intensive. Here we propose a novel generative model: Convolutional Autoencoder Model to add unsupervised adversarial training, i.e., the production of adversarial...
Many practical machine learning tasks employ very deep convolutional neural networks. Such large depths pose formidable computational challenges in training and operating the network. It is therefore important to understand how many layers are actually needed to have most of the input signal's features be contained in the feature vector generated by the network. This question can be formalized by...
Smart vision systems on a chip are promising for embedded applications. Currently, flexibility in the choice of integrated pre-processing tools is obtained at the expense of total silicon area and fill factor, which are otherwise optimized provided that the sensor performs a specific task. We propose a new architecture based on macropixel-level processing to improve the trade-off by using the same...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.