Search

Due to the influence of noise in the received signal in non-cooperative communication, it is difficult for existing Automatic modulation classification methods to balance classification accuracy and model complexity. This paper proposes a novel Convolutional Adaptive Noise Reduction (CANR) network, which consists of an Adaptive Noise Reduction (ANR) module and a Convolutional Feature Extraction (CFE) module. The ANR and CFE modules denoise the combined input and capture the spatiotemporal features in the time series. Experiments on benchmark datasets show that the proposed network has the fewest training parameters and state-of-the-art recognition accuracy under the same conditions.

Mobile crowd computing (MCC) that utilizes public-owned (crowd’s) smart mobile devices (SMDs) collectively can give adequate computing power without any additional financial and ecological cost. However, the major challenge is to cope with the mobility (or availability) issue of SMDs. User’s unpredicted mobility makes the SMDs really unstable resources. Selecting such erratic resources for job schedule would result in frequent job offloading and, in the worst case, job loss, which would affect the overall performance and the quality of service of MCC. In a Local MCC, generally, a set of users are available for a certain period regularly. Based on this information, the chances of a user being available for a certain duration from a given point of time can be predicted. In this paper, we provide an effective model to predict the availability of the users (i.e., their SMDs) in such an MCC environment. We argue that before submitting a job to an SMD, the stability of it is to be assessed for the duration of execution of the job to be assigned. If the predicted availability period is greater than the job size, then only the job should be assigned to the SMD. An accurate prediction will minimize the unnecessary job offloading or job loss due to the early departure of the designated SMD. We propose an advanced convolutional feature extraction mechanism that is applied to LSTM and GRU-based time-series prediction models for predicting SMD availability. To collect user mobility data, we considered a research lab scenario, where real mobility traces were recorded with respect to a Wi-Fi access point. We compared the prediction performances of convolutional LSTM and GRU with the basic LSTM and GRU and ARIMA in terms of MAE, RMSE, R2, accuracy, and perplexity. In all the measurements, the proposed convolutional LSTM exhibited considerably better prediction performance.

This paper aims at the monitoring of birds and their ecological environment in the island and coastal wetland ecosystems. A new approach of island bird detection is proposed based on the Faster R-CNN (Regions with Convolutional Neural Networks) model under multiple backgrounds. It includes feature extraction, region proposal, bounding box regression, classification into the whole neural network structure. This key technology can automatically achieve automatic bird species identification and quantitative statistics in the faster computation speed. The details of constructing Faster R-CNN are described. In the end, many actual images are utilized to demonstrate the effectiveness of the proposed models. [ABSTRACT FROM AUTHOR]

Abstract: This work presents a holistic system for the off-line recognition of cursive words based on the extraction of perceptual features by means of convolution with templates of line segments. The method works directly on gray-level images to avoid losing information due to binarization. Features are represented spatially in order to preserve topological information which is well suited for recognition through convolutional neural networks. Moreover, our feature extraction method does not need to detect baselines to find ascenders and descenders. The system has been tested on small and medium vocabulary sizes demonstrating the feasibility of the proposed method. [Copyright &y& Elsevier]