A physical method is developed to analyze the Riemann zeta function and the L-function. First, a physical model of an elastic bar fixed at both ends and subjected to symmetric axial loads with respect to the mid-span is constructed. The axial force of the bar at the left end is calculated with two methods: the equilibrium of forces and the principle of minimum potential energy. With the help of the orthogonality of {sinix} (i 1, 2, 3, ) , the Parseval identity and the Bessel inequality of the physical model are obtained. Further, it is proven that, in all the possible displacements which satisfy the boundary conditions, the real one minimizes the total potential energy of the bar. With the equivalence of the two methods, an identity of a type of infinite series is derived. Based on the identity, L(1) , L(3) , a recurrence formula of L(2k 1) (k  3) ,  (2) ,  (4) , and a relationship between  (2k) (k  2) and L(2i 1) (i  1, 2, ..., k) are then deduced with power axial load functions. The upper and lower bounds for the Riemann zeta function  (2k -1) and the L-function L(2k) are given. The numerical results show that the upper and lower bounds are in excellent agreement with the accurate value. Finally, two improvement methods are proposed for estimating the circumference ratio. The numerical results show that the two methods can estimate the circumference ratio with high accuracy and that the L -function is more efficient than the Riemann zeta function in estimating the circumference ratio.

Nowadays, pictures have more and more applications in various fields. Because different situations may have different commands of the ratios of pictures, we present a program which can resize the pictures according to their contents. Compared with the traditional method which only cares about the geometric constraints, our program can reserve the important contents of the pictures, which means that the visual effect of the pictures will not be damaged after the resizing. In order to change the ratio of pictures while avoiding distorting their main contents, we use a method called seam carving. Based on the energy function used for finding useless pixels and dynamic programming used for finding some continuous seams which will be deleted later, our program can finally change the image to a new ratio perfectly. Due to some deficiencies of the basic program, we also make some improvements. The first one is the improvement of the algorithm: we add face detection to the basic program. Not only does this improvement can solve many problems of the basic program automatically, but also it is our own innovation of the study of content-aware resizing. The second one is the addition of customization: we allow users to select the specific contents that they want to reserve, which can meet the different needs of users. It’s also our own innovation which will be very useful in practical situations. Our program can have many practical applications. We have some ideas about how to put this algorithm into real use, which is one of our practical innovations of this topic. In terms of different ratios of photo printing, graphic designing and screen displaying, this program will have various uses, which will make a groundbreaking progress in practice.

Information Technology has enabled new levels of convenience to our daily life. While providing people with astonishingly good experiences, computer science has still a long way to meet the needs from the users for various scenarios. As a good example, path selection problem is an important and common problem which finds applications in location based services, internet routing and navigation systems. Choosing the most cost effective and time saving routines plays an essential role in both theory and engineering. Although a plethora of classical solutions have been made to the static shortest path problem, dynamic shortest path problem has not been thoroughly studied. To address this critical issue, this paper investigates the dynamic shortest path problem by optimizing existing work and conducting extensive experiments. The studies fall into three main variants of the problem: unweighted shortest paths, weighted shortest paths, and shortest paths on the map. We present a comprehensive comparison of our optimizations and existing solutions which demonstrates the effectiveness and efficiency of our algorithms.

Opinions and sentiments are essential to human activities and have a wide variety of applications. As many decision makers turn to social media due to large volume of opinion data available, efficient and accurate sentiment analysis is necessary to extract those data. Hence, text sentiment analysis has recently become a popular field and has attracted many researchers. However, extracting sentiments from audio speech remains a challenge. This project explored the possibility of applying supervised Machine Learning in recognizing sentiments in English utterances on a sentence level. In addition, the project also aimed to examine the effect of combining acoustic and linguistic features on classification accuracy. Six audio tracks were randomly selected to be training data from 40 YouTube videos (monologue) with strong presence of sentiments. Speakers expressed sentiments towards products, films, or political events. These sentiments were manually labelled as negative and positive based on independent judgement of 3 experimenters. A wide range of acoustic and linguistic features were then analyzed and extracted using sound editing and text mining tools respectively. A novel approach was proposed, which used a simplified sentiment score to integrate linguistic features and estimate sentiment valence. This approach improved negation analysis and hence increased overall accuracy. Results showed that when both linguistic and acoustic features were used, accuracy of sentiment recognition improved significantly, and that excellent prediction was achieved when the four classifiers were trained respectively, namely kNN, SVM, Neural Network, and Naïve Bayes. Possible sources of error and inherent challenges of audio sentiment analysis were discussed to provide potential directions for future research.

Nowadays, China has become the country with the largest aging population. The increasing proportion of the elders inspires the development of related industries. Falling is extremely dangerous for the elders. Thus, it is of vital importance to detect it and send an SOS signal immediately when an elder falls. Existing methods in fall detection include vision based, wearable sensor-based, ambinent device-based techniques. However, there are drawbacks in these existing methods. The video detection cannot guarantee the privacy of the elderly. Monitoring the changing of the environment requires high cost and its accuracy may be degraded by similar phenomena like strenuous exercise or a heavy object dropping down. The wearable sensors need to be worn all the time which is uncomfortable. In this paper, widely deployed Wi-Fi system is used for detecting a fall. It has little influence on the daily life of the elderly and can make up the drawbacks of the methods above. Because human movements would affect wifi signal propagation, we could detect human behaviors, like falling, through wifi. We firstly extract Channel State Information(CSI) from the original wifi signals collected. After performing noise and dimensionality reduction via Principal Component Analysis (PCA), the characteristic data of the Wi-Fi signal is obtained. Then, the detection model is trained using Recurrent Neural Networks (RNN) based on the PyTorch deep learning. Finally, fall detection is achieved automatically based on the trained model. For further improvement of its function, a strategy based on Random State Reset (RSR) and RNN is brought up in this paper, called RSR-RNN. Related experiments have been conducted to verify the feasibility of the method developed in this paper. Six types of fall scenarios are designed according to the data collected under different situations. The results of the experiments show that the RNN model can detect a fall effectively at an accuracy of 84.67%. The average accuracy of detection is further increased to 85.83% using RSR-RNN, which indicates that this strategy is effective. Keywords:

Clothing is one of the symbol of modern fashion trends, research concerning clothing detail’s recognition still remains a hot topic in the field of computer. Existing research is mainly focused on relating clothing detail features and aesthetic styles, realizing the possibilities of auto clothing aesthetic appreciation. In this paper, based on existing data sets, we will introduce the ideology of Deep Learning into the association model between clothing visual features to improve the model’s effectiveness and accuracy. By calculating the distance between aesthetic words and seed words, we are able to map the most often used aesthetic words in a two-dimensional space, building a fashion semantic space. Later, we introduce the ideology of Deep Learning into the association model between clothing detail features, and designed experiments to prove the effectiveness of the model. In the research, we discussed how different parameters and structures in Deep Learning effect the outcome of the relating model, and by comparison we choose the best parameters of the model. Based on the work mentioned above, we analyze different styles of different categories’ clothing in shopping websites, which further confirms the effectiveness of our model.

With the rapid development of infrastructure including power grids, managers in power industry these days are faced with an increasingly severe problem of theft of electricity. Theft of electricity has negative effects on many socioeconomic aspects, including impacting stable growth of economic for power enterprises and social development. The traditional anti-theft means require officers checking the integrity of kilowatt-hour meter and the correctness of wiring house by house, which requires enormous manpower and material resources. With the advancement of information collecting technology, power enterprises now possess relatively complete database of power consumption. As a result, performing data mining on existing database and identifying abnormal users has become a hot topic in the field of information technology. The purpose of this paper is to identify abnormal users with machine learning algorithms, providing power enterprises with means to detect theft of electricity at lower cost. The contributions of this paper are listed as follows: 1) Propose various features, providing means to describe users’ behaviors. First, monistic features (mean, difference, coefficient of variation, range, standard deviation), binary features (cosine similarity, Pearson product-moment correlation coefficient) and multivariate features are calculated based on user power consumption (monthly, seasonally, yearly, per holiday, per workday). Then principal component analysis is used to perform dimensionality reduction on the dataset. The dataset is finally scaled and oversampled to form the feature dataset. 2) Study various classification algorithms, optimize hyperparameters, providing models to detect theft of electricity. In this paper, the mechanism behind support vector machine, back-propagation neural network, random forest and XGBoost is studied and the hyperparameters are optimized. 3) Compare and evaluate different classifiers, and provide suggestions for real-world application. In this paper, different classifiers are evaluated with different metrics and the best classifier is recommended based on real-world dataset and different application scenarios.

This work focuses on the problem of high quality style transfer for Chinese characters, converting Chinese printed with standard typograph into those in calligraphic style. Traditional approaches include employing derivative network structures of CNN, RNN, and GAN, yet generation results are not satisfactory enough. In this paper, we proposed an integrated system of pix2pix structure, auxiliary classifier, and WGAN. Trials were run to evaluate the performance of traditional CNN and GAN. Cross-comparative sets of experiments among GAN, AC-GAN, and our proposed AC-WGAN have been conducted to test the capabilities of our proposed model. Inference and interpolation tests are conducted as well. Our proposed model outperformed existing style transfer system in generation’s delicacy, similarity, and efficiency. Incorporated with WGAN, the model also demonstrated a strong ability in providing a truthful training indicator with its loss. Study also suggested that AC and pix2pix adaption hold huge significance in accelerating the training process. Research validates the viability of fulfilling Chinese character transformation with style transfer. With handwriting recognition network, fast, high-quality Chinese character style transfer forms the basis for instantaneous conversion between printed and written work. Optimization on algorithm may be another direction for further exploration.

▲With the rapid development of bicycle-sharing system since 2016, illegal parking has become one of the most significant social issues. While the existing management methods cannot meet the demand, object detection technology provides a possibility. The Video Surveillance System can be utilized as a rich source of information. However, there is technical challenge to handle complex scenarios. Faster R-CNN, chosen because of its excellent overall performance, cannot specify the proposals of shared bikes especially in distant view. Processing the consecutive images in videos is neither necessary nor efficient. The original results are limited by interference of moving bikes and blocking effects. ▲ In this paper, I introduce a Faster R-CNN over Attention (FoA) that combines the from-coarse-to-fine visual mechanism with deep neural networks. FoA realizes fine-grained detection of shared bikes in complex circumstances. It consists of an Attention Region Extraction (ARE) and an optimized Faster R-CNN. In ARE, I introduce the concept of Attention Region and define it as video background, then apply Gaussian Mixture Model. ARE processes the surveillance video into discrete regions, allowing concentration on certain areas and efficiency in detection. In Faster R-CNN, I put forward an anchor box optimization in regards of the remarkable region-based characteristic in R-CNN. The optimization applies k-means to refine the anchor boxes and generate more appropriate region proposals. ▲ I construct a categorized shared bike data set of 4,291 images and 12,697 labeled objects for FoA training and testing. FoA performs excellently with the detection and recognition ratios of 94.19% and 92.73% respectively. The optimization is proved to make a difference. ▲ The paper not only provides a new direction for solving illegal parking issue and managing bicycle-sharing system, but also generalizes FoA into a new detection framework of Attention Region Extraction plus Region-Based Convolutional Neural Network, which can be applied to certain object detection in various scenes. The mechanism that combines ARE and R-CNN organically allows the towards-real-time detection to possess both practicality and universality. FoA is one of its robust implements.

3D shape classification plays a fundamental role in geometric big data analysis. This work proposes a novel method for shape classification based on optimal mass transportation theory. The Riemann surfaces are mapped onto canonical domains conformally based on uniformization theorem. The conformal factor function is treated as probability distribution on the canonical domain. For each pair of probability distributions, the optimal mass transportation map is computed by solving Monge-Amper´e equation. The transportation cost is theWasserstein distance between two distributions. By using this distance, geometric classification based on clustering can be performed. The method is applied to 3D human facial expression recognition, which demonstrates the efficiency and efficacy of the method.