T R A C K       P A P E R
Close

Login Panel

Close tab

Password Meter

Volume 07 Issue 10 (October 2020)

S.No. Title & Authors Page No View
1

Title : Nitrification and Denitrification Process Employing Three Different Sunken Materials Types in Biological Aerated Filter (BAFs)

Authors : Adel S. Ibrahim Faskol, Gabriel Racovițeanu

Click Here For Abstract

Download Certificate
Abstract :

The overall aim of this research was to use easily obtained materials as an attached growth zone in the treatment of municipal wastewater. Where investigate their performance in the nitrification and denitrification process. were using three identical pilot-scale reactors of the biological aerated filter system BAFs, down-flow mode. Operated at ambient air temperature ranged from 8 ◦C to 29◦C with a mean 17.93 ±7.27 ˚C. While the hydraulic retention time HRT was 12-hours and the influent was 100% recirculation, with daily backwashes. As results of the experiments showed that the greatest mean removal efficiency of the NH4+-N was in the reactor employing activated carbon-based material bed where achieved 90.11%. And then 87.74%, 85.17% respectively for the reactor employing sand-based material bed, and the reactor employing ceramic particle-based material bed. In addition, the pilot-scale reactors of the BAFs were able to nitrify between 0.23±0.22 Kg NH4+-N/m3/day to 0.24±0.21 Kg NH4+-N/m3/day

1-5
2

Title : A Survey of Semi-supervised Learning Research

Authors : Yapei Zhao

Click Here For Abstract

Download Certificate
Abstract :

With the rapid growth of modern data sets and increasingly passive data collection, the cost of tagged data is becoming higher and higher, and unlabeled data is not only cheap but sufficient in many applications. The use of unlabeled data to improve the prediction of machine learning systems is a semi-supervised learning problem (SSL). Semi-supervised learning has high practical application value and can improve the learning performance of the model. This article will describe the semi-supervised learning method in three parts. Firstly, it describes the definition and development process of semi-supervised learning, then describes the commonly used algorithms and practical applications of semi-supervised learning, and finally discusses the future research direction of semi-supervised learning

6-8
3

Title : Research on visual Question answering System Method based on deep learning

Authors : An Chang, Anna Wang

Click Here For Abstract

Download Certificate
Abstract :

Visual Questions and Answers (VQA) is a new research task combining computer vision (CV) and natural language processing [1]. Given an image and a natural language problem, it needs to combine the information of the two modes, that is, according to the visual features of the image and the text features of the problem. And then combine the information of the two features to get the right answer. Firstly, the feature extraction method and feature fusion method of visual and text are introduced. Secondly, the effect of attention mechanism on visual question-answering model is introduced. The data set used to train and evaluate the visual question answering system was then reviewed. Finally, we discuss the future direction of the field

9-13
4

Title : A Survey on Research Progress of Generative Adversarial Networks

Authors : Yan Bai

Click Here For Abstract

Download Certificate
Abstract :

Purpose Since the birth of Generative Adversarial Networks (GANs), it has become a research hotspot in the field of machine learning and artificial intelligence. GAN uses the principle of adversarial training between the generator and the discriminator to solve many problems that are difficult to solve by traditional models. Based on the advantages of GAN, more and more people have begun to conduct in-depth research on it, resulting in many variants of GAN. With the continuous improvement of GAN, it has played a surprising role in some application fields, such as the visual field, image field, audio field, natural language processing field and various other fields. This article first introduces the GAN model and its basic principles, then introduces the popular variant models of GAN, and finally summarizes the application status and research progress of GAN in various fields

14-15
5

Title : Image in Painting through Edge Prediction Generator

Authors : Jialiang Yan

Click Here For Abstract

Download Certificate
Abstract :

In the past few years, deep learning technology has significantly improved image restoration. However, many of these techniques cannot reconstruct reasonable structures because they are usually too smooth and/or blurry. This article develops a new image restoration method that can better reproduce the exquisite details of the filled area. We propose a two-stage adversarial GAN network that includes an edge generator, followed by an image completion network. The edge generator makes the edges of the missing areas (regular and irregular) of the image illusion, and the image completion network uses the illusion to fill the missing areas of the image
Priori edge. We conducted an end-to-end evaluation of the publicly available datasets CelebA, Places2, and showed that it is superior in quantity and quality to the current state-of-the-art technology

16-18