12th International Conference on Computer Science, Engineering and Information Technology (CCSEIT 2022)

February 26 ~ 27, 2022, Vancouver, Canada

Accepted Papers


MultiDTA: Drug-target Binding Affinity Prediction Via Representation Learning and Graph Convolutional Neural Networks

Deng Jiejina, Zhang Yijiaa, Dong Xinruia, Zhang Fuyaoa and Lu Mingyua, School of Information Science and Technology, Dalian Maritime University, Dalian 116024, Liaoning, China

ABSTRACT

Predicting drug-target interaction (DTI) plays a crucial role in the study of drug repositioning. In computer-aided drug development, high-performance computers are used to simulate drug development tasks, which is a promising area of research. In drug target affinity prediction, compared to several statistical and machine learning-based models that have been proposed, the deep learning approach is better than the traditional methods. In order to improve the accuracy, the prediction of drug-target affinity (DTA) based on deep learning has always been the focus of research. More and more advanced models have been proposed in recent years, but they ignore the information of different forms of data and do not make full use of data information. This paper proposes a novel end-to-end learning framework for DTA prediction called MultiDTA. In this model, we use multi-channel inputs to predict drug-target affinity, which can make full use of the different information of the data. Specifically, Extraction of sequence and contextual information of drugs and proteins using convolutional neural networks and LSTM. The drug and protein sequences convert into graph structures, then using graph neural networks to extract spatial structure information from drug and protein. We conduct extensive experiments to compare our proposed with state-of-the-art models. Our model is highly competitive relative to other models. The code of MultiDTA and the relevant data are available at: https://github.com/dengjiejin/MultiDTA.

KEYWORDS

drug-target affinity prediction, representation learning, multi-channel inputs, GCN.


A Recursive Threshold Secret Sharing Scheme for Cloud Storage

Huizhu Li and Lixi Fang, School of Information, Central University of Finance and Economics, Beijing, 102206

ABSTRACT

In recent years, cloud computing has received extensive attention from everyone. Cloud storage is one of the most widely used services in cloud computing, with a wide range of applications, but some security issues have followed. One of the more serious security issues is that user privacy data is leaked or modified by cloud storage service providers, causing serious losses to users. In response to this problem, this paper proposes a recursive polynomial secret sharing threshold scheme in which users and cloud storage service providers jointly manage user privacy data. In this scheme, the last two polynomials in the recursive equation are used to distribute secret shares to cloud storage service providers and users respectively, so that only users and cloud storage service providers that exceed the threshold can cooperate to recover the key. Then the paper analyzes the safety of the scheme, mainly analyzing whether the scheme is safe from three aspects: correctness, computational safety and robustness. Finally, the paper uses a simple example to show that the scheme can effectively protect the confidentiality of users private data.

KEYWORDS

Cloud Storage, Threshold Secret Sharing, Recursive Secret Sharing, Cloud Storage Service Provider.


Artificial Intelligence Systems for Teaching and Learning in Civil Engineering: Conceptual Framework

L O Toriola-Coker1*, Nureni Asafe Yekini1, H Alaka2, 1School of Engineering, Yaba College of Technology, Yaba Lagos, 2University of Hertfordshire, Hatfield, Hertfordshire, UK

ABSTRACT

Artificial intelligence technology is based on design of machine or computer application that mimic human intelligent. Use of artificial intelligence in teaching and learning in civil engineering is a welcome development. This paper presents a conceptual framework of Artificial Intelligence Systems for Teaching, Learning, and administration of in Civil Engineering education. The proposed system is to be design using the following tools: Extensible Markup Language (XML) to develop the GUI, Hypertext Preprocessor (PHP) for the web user interface (WUI), APACHE for middleware, MYSQL for database design, and UML will be used to visualize the design of the system. If system is developed and implemented it will go a long way to advance teaching and learning, and educational administration in civil engineering profession.

KEYWORDS

Artificial Intelligence, learning management system, Civil Engineering, MYSQL.


Lightblue: Nurture Your Personal Chatbot

Xiang Zhang1, Yan Liu2, Gong Chen3 and Sheng-hua Zhong4, 1Department of Computing, The Hong Kong Polytechnic University, Hong Kong, China, 2Department of Computing, The Hong Kong Polytechnic University, Hong Kong, China, 3Department of Computing, The Hong Kong Polytechnic University, Hong Kong, China, 4College of Computer Science and Software Engineering, Shenzhen University, Shenzhen, China

ABSTRACT

Chatbot has always been an important research topic in artificial intelligence and attracts lots of attention recently. Despite the great progress in language ability, the interactions between users and chatbots are rather generic, short-term, and transnational. It has always been challenging to develop truly personal chatbots and even establish long-term and affective connection. This paper first brings up “nurture” as a new interaction mode with chatbots. We introduce the nurture framework and design the learning algorithm and nurture functions accordingly. Then we present LightBlue – a platform that allows non-professionals to nurture personal chatbots from scratch. Experiments on both close-domain and open-domain tasks have showed the effectiveness of the proposed framework and demonstrated a promising way to establish a longterm interaction between users and chatbots.

KEYWORDS

Personal Chatbot, Conversation Agent, Nurture, Human-chatbot Interaction, Long-term Relationship.


An Automated Video Content Customization System using Eye Tracking and Artificial Intelligence

Yuyang Lou1, Yu Sun2, 1Charles Wright Academy, 7723 Chambers Creek Rd W, Tacoma, WA 98467, 2California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

In the past few years, the internet and online social networks developed drastically, promoting the development of online learning programs [1][2]. These programs provided opportunities for a digital learning experience that allows students to explore beyond whats taught in school. However, having a clear understanding of what topic might interest the user and motivate the user to further explore that topic is hard for both the user and the learning program. This paper proposes to create one centralized method of predicting what the user would be interested in and provide them with educational content recommendations. Our design builds upon the eye-tracking techniques, which allows us to capture users’ eye movements, and object recognition achieved by machine learning, which allows us to examine the specific object that the users are looking at and provide data for the users’ interest analysis [3]. Our results show a success rate of __% of analyzing what the user is truly looking at. We used our decision heuristic, etc.

KEYWORDS

Eye Tracking, Deep learning, computer vision.


Trusting Machine Learning Algorithms in Predicting Malicious Nodes Attacks

Basim Mahbooba, Mohan Timilsina And Martin Serrano, The Insight centre, NUIG University, Galway City, Ireland

ABSTRACT

Identifying network attacks is a very crucial task for Internet of things (IoT) security. The increasing amount of IoT devices is creating a massive amount of data and opening new security vulnerabilities that malicious users can exploit to gain access. Recently, the research community in IoT Security has been using a data-driven approach to detect anomaly, intrusion, and cyber-attacks. However, getting accurate IoT attack data is time-consuming and expensive. On the other hand, evaluating complex security systems requires costly and sophisticated modelling practices with expert security professionals. Thus, we have used simulated datasets to create different possible scenarios for IoT data labeled with malicious and non-malicious nodes. For each scenario, we tested off a shelf machine learning algorithm for malicious node detection. Experiments on the scenarios demonstrate the benefits of the simulated datasets to assess the performance of the ML algorithms.

KEYWORDS

IoT Simulation, Data Labels, Malicious Nodes, Attacks, Trust, Prediction.


iOLLVM: Enhanced Version of OLLVM

Chengyang Li, Tianbo Huang, Xiarun Chen, Chenglin Xie, Weiping Wen, School of Software and Microelectronics, Peking University, Beijing, China

ABSTRACT

Code obfuscation increases the difficulty of understanding programs, improves software security, and, in particular, OLLVM offers the possibility of cross-platform code obfuscation. For OLLVM, we provide enhanced solutions for control flow obfuscation and identifier obfuscation. First, we propose the nested switch obfuscation scheme and the in-degree obfuscation for bogus blocks in the control flow obfuscation. Secondly, the identifier obfuscation scheme is presented in the LLVM layer to fill the gap of OLLVM at this level. Finally, we experimentally verify the enhancement effect of the control flow method and the identifier obfuscation effect and prove that the programs security can be further improved with less overhead, providing higher software security.

KEYWORDS

Software Protection, Code Obfuscation, Control Flow Obfuscation, Identifier Obfuscation, LLVM.


Speech To Text Analysis or: How I Learned to Stop Worrying and Love Academic Articles

Isaac Tijerina1 and Dr. Soma Datta2, 1Masters of Science in Software Engineering Student, University of Houston – Clear Lake, Houston, Texas, USA, 2Associate Professor of Software Engineering, University of Houston – Clear Lake, Houston, Texas, USA

ABSTRACT

Objective: The focus of this study is to research what has already been accomplished in Speech to Text with usage in programming and the general application of Speech to Text in other fields. What is found can then be applied to furthering the application of Speech to Text with coding. Results: It was found that the state of modern Speech to Text is in constant motion. There is work being done in the field constantly to improve Speech to Text and apply it to various fields. In its usage, it is being applied to medical fields, education, machinery control, and many others. It is being seen that while being used there is still a struggle with a users accent if it is different from the native accent of the language.

KEYWORDS

Speech to text, computer speech recognition, speech to text accent, spoken word recognition.


An Intelligent Mobile Application for Car License Plate Detection And Analysis Using Machine Learning Algorithm

Mengcheng Han1 and Yu Sun2, 1Santa Margarita Catholic High School, 22062 Antonio Pkwy, Rancho Santa Margarita, CA 92688, 2California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

How did image recognition and object analysis function to bring convenience to people’s lives? Within this question I bear in my mind, I started to explore and build this Automatic Car License Plate Detect and Analyze Project. Since cameras are being widely used for recording and analyzing vehicle information, it has been a great cost to buy such intelligent devices. Guided by recent research on machine learning approaches , we solve this financial problem by designing and implementing a mobile phone application that automatically utilizes the camera installed on the phone to analyze the information of the car license plate. Our design is built to provide users with an accessible way to analyze license plates in complex environments.

KEYWORDS

Machine learning, LPR, OCR libraries.