Recently, deep learning methods, mostly algorithms based on Deep Convolutional Neural Networks (DCNNs) have yielded great results on pedestrian detection. Algorithms based on DCNNs spontaneously learn features in a supervised manner and are able to learn qualified high level feature representations to detect pedestrian. In this paper, we first review a number of popular DCNN-based training approaches along with their recent extensions. We then briefly describe recent algorithms based on these approaches. Also, we accentuate recent contributions and main challenges of DCNNs in detecting pedestrian. We analyze deep pedestrian detection algorithms from training approach, categorization, and DCNN model points of view, and ultimately propose a new deep architecture and training approach for deep pedestrian detection. The experimental results show that the proposed DCNN and training approach, achieve more accurate rate detection than the previously reported architectures and training approaches.

According to world health organization, breast cancer is one of the most deadly cancers occurred in women. Therefore accurate diagnosis and prediction is important to decrease the high death rate. The aim of this paper is twofold. First, improving breast cancer detection accuracy using Modified Fuzzy Logic (MFL) then improving the performance of MFL algorithm using GPU platform. The experimental results show that the accuracy of the breast cancer detection using MFL is higher than other techniques. In addition, by exploiting loop-level parallelism and pipeline parallel communication pattern in MFL algorithm, its performance is improved up to 19.17x for different image sizes.

This paper develops an automatic parking algorithm based on a fuzzy logic controller with the vehicle pose for the input and the steering angle for the output. In this way some feasible reference trajectory path have been introduced according to geometric and kinematic constraints and nonholonomic constraints to simulate motion path of car. Also a novel method is used for parking space detection according to image processing. A fuzzy controller according to experiments of skilled driver and path planning is designed, and then fuzzy rules are tuned and finally fuzzy membership functions are optimized using genetic algorithm. Simulation results illustrate the effectiveness of the developed schemes.

Background and Objectives: Louvain is a time-consuming community detection algorithm especially in large-scale networks. Using Graphic Processing Unit (GPU) in order to calculate modularity sigma, which is a major processing section in Louvain algorithm, can reduce algorithm execution time and make it practical for large-scale networks. Methods: The proposed algorithm Dynamic CUDA Louvain Method (DCLM) blocks hardware threads dynamically on cores inside GPU. By considering the properties of GPU, this algorithm allocates the maximal number of processing cores to each Stream Multi-Processor (SM) as number of threads in a block. If the number of nodes in the graph is smaller than all physical cores on GPU, number of threads per block Is equal to the ratio number of graph nodes over the number of SMs. Results: The implementation results demonstrated that the proposed algorithm is able to decrease the run time by 15% in comparison with the best past method in the large-scale graph. Conclusion: We have introduced DCLM algorithm based on GPU that accelerates Louvain community detection algorithm. Dynamic allocation of threads to each block has a significant effect on the reduction of algorithm execution time. However, incrementing the number of threads per block alone does not result to acceleration the speed of calculations.

Nowadays، several infrastructure-based low-frequency acoustical sensor networks are employed in different applications to monitor the activity of diverse natural and man-made phenomena، such as avalanches، earthquakes، volcanic eruptions، severe storms، super-sonic aircraft flights، etc. Two signal detection methods are usually implemented in these networks for the purpose of event occurrence identification، which are the progressive multi-channel correlator (PMCC) and the so-called Fisher detector. But، the Fisher method is more important and applicable in low signal-to-noise (SNR) ratio conditions، which is of a special interest in acoustical monitoring networks. Unfortunately، an important disadvantage of this algorithm is its relative high detection-time; which limits its application for real-time detection scenarios. This disadvantage is fundamentally due to a beam forming process in Fisher algorithm، which requires doing complete search in a slowness-network، constructed from possible incoming wave front directions and speeds. To address this issue، we propose a method for implementation of this beam forming on a graphics processing unit (GPU)، in order to realize a fast-computing and/or near real-time signal processing technique. In addition، we also propose a parallel-processing algorithm for further enhancement of the performance of this GPU-based Fisher detector. Simulation results confirm the performance improvement of Fisher detector، in terms of required processing time for acoustical signal detection applications.

In parallel to the increasing use of electronic cards, especially in the banking industry, the volume of transactions using these cards has grown rapidly. Moreover, the financial nature of these cards has led to the desirability of fraud in this area. The present study with Map Reduce approach and parallel processing, applied the Kohonen neural network model to detect abnormalities in bank card transactions. For this purpose, firstly it was proposed to classify all transactions into the fraudulent and legal which showed better performance compared with other methods. In the next step, we transformed the Kohonen model into the form of parallel task which demonstrated appropriate performance in terms of time, as expected to be well implemented in transactions with Big Data assumptions.

Milk and dairy products as an animal protein source with a high nutritional value has a particular importance in human diet. Development of non-destructive and quick methods for assessing the quality and freshness of milk is of particular importance in ensuring product safety before consumption. The aim of this study was to measure the dielectric power with the use of a parallel-plate sensor to develop a new method for rapid assessment of milk freshness. Three milks with commercial names of Pak, Kaleh and Damdaran and a milk from dairy farm were dielectrically measured during seven storage days. Sinusoidal voltage wad applied to the milk samples and capacitor power within the frequency range of 0-150 MHz was measured using a spectrum analyzer. Results proved that the frequency range of 75-100 MHz varied considerably with storage time so that the frequency of the valley within this frequency range increased with storage time. The frequency of the valley as a spectrum characteristic predicted the storage day by a linear regression model with R2 and RMSE of 0.84 and 0.8, respectively. The correlation of milk pH and the frequency of the valley showed that during the first four days of storage with a minor change in pH, the frequency of valley varied significantly indicating further physico-chemical changes in milk with storage that affects the dielectric properties of milk. The dielectric power at the frequencies of 2.7 and 89.4 as the most effective frequencies, predicted the storage day and pH, respectively with RMSEs of 0.4 and 0.22. Multiple linear regression models decreased the errors of prediction to 0.26 and 0.11 for storage day and pH, respectively. Based on the findings of this study, dielectric power spectroscopy can be implemented as a non-destructive, simple and accurate method to milk pH and freshness determination.

Automatic topic detection seems unavoidable in social media analysis due to big text data which their users generate. Clustering-based methods are one of the most important and up-to-date categories in topic detection. The goal of this research is to have a wide study on this category. Therefore, this paper aims to study the main components of clustering-based-topic-detection, which are embedding methods, distance metrics, and clustering algorithms. Transfer learning and consequently pretrained language models and word embeddings have been considered in recent years. Regarding the importance of embedding methods, the efficiency of five new embedding methods, from earlier to recent ones, are compared in this paper. To conduct our study, two commonly used distance metrics, in addition to five important clustering algorithms in the field of topic detection, are implemented by the authors. As COVID-19 has turned into a hot trending topic on social networks in recent years, a dataset including one-month tweets collected with COVID-19-related hashtags is used for this study. More than 7500 experiments are performed to determine tunable parameters. Then all combinations of embedding methods, distance metrics and clustering algorithms (50 combinations) are evaluated using Silhouette metric. Results show that T5 strongly outperforms other embedding methods, cosine distance is weakly better than other distance metrics, and DBSCAN is superior to other clustering algorithms.