keyboard_arrow_up
Accepted Papers
Research on Task Scheduling Strategy Based on the Trustworthiness of MapReduce

QIN Jun1,2, Song YanYan1 and ZONG Ping2,3, 1Communication University of China, Nanjing, 2Nanjing University of Posts and Telecommunications, Nanjing, China, 3Nanjing University of Science and Technology Zijin Colleg, China

ABSTRACT

With the rapid development and popularization of information technology, cloud computing technology provides a good environment for solving massive data processing. Hadoop is an open-source implementation of MapReduce and has the ability to process large amounts of data. Aiming at the shortcomings of the fault-tolerant technology in the MapReduce programming model, this paper proposes a reliability task scheduling strategy that introduces a failure recovery mechanism, evaluates the trustworthiness of resource nodes in the cloud environment, establishes a trustworthiness model, and avoids task allocation to low reliability node, causing the task to be re-executed, wasting time and resources. Finally, the simulation platform CloudSim verifies the validity and stability of the task scheduling algorithm and scheduling model proposed in this paper.

KEYWORDS

Cloud environment, Failure recovery mechanism, Task scheduling algorithm.


A Fragmentation Region-Based Skyline Computation Framework for a Group of Users

Ghoncheh Babanejad Dehaki1, Hamidah Ibrahim1, Nur Izura Udzir1, Fatimah Sidi1 and Ali Amer Alwan2, 1Universiti Putra Malaysia, Selangor, Malaysia, 2International Islamic University Malaysia, Kuala Lumpur, Malaysia

ABSTRACT

Skyline processing, an established preference evaluation technique, aims at discovering the best, most preferred objects, i.e. those that are not dominated by other objects, in satisfying the user’s preferences. In today’s society, due to the advancement of technology, ad-hoc meetings or impromptu gathering are becoming more and more common. Thus, identifying objects within a predetermined region that best meet the preferences of a group of users is inevitable. Although there are works that have been conducted on this aspect, computing the skylines for various groups of users in similar region would mean rescanning the objects of the region and repeating the process of pairwise comparisons among the objects which are undoubtedly unwise. On this account, this study presents a region-based skyline computation framework which attempts to resolve the above issues by fragmenting the search region of a group of users and utilising the past computed skyline results of the fragments. The skylines, which are the objects recommended to be visited by a group of users, are derived by analysing both the locations of the users, i.e. spatial attributes, as well as the spatial and non-spatial attributes of the objects. Several experiments have been conducted and the results show that our proposed framework outperforms the previous works with respect to CPU time.

KEYWORDS

Skyline Queries, Preference Queries, Group Preferences, Fragmentation Strategy.


How Many Features is an Image Worth? Multi-channel CNN for Steering Angle Prediction in Autonomous Vehicles

Jason Munger and Carlos W. Morato, Department of Robotics Engineering, Worcester Polytechnic Institute, Worcester, MA

ABSTRACT

This project explores how raw image data obtained from AV cameras can provide a model with more spatial information than can be learned from simple RGB images alone. This paper leverages the advances of deep neural networks to demonstrate steering angle predictions of autonomous vehicles through an end-to-end multi-channel CNN model using only the image data provided from an onboard camera. Image data is processed through existing neural networks to provide pixel segmentation and depth estimates and input to a new neural network along with the raw input image to provide enhanced feature signals from the environment. Various input combinations of Multi-Channel CNNs are evaluated, and their effectiveness is compared to single CNN networks using the individual data inputs. The model with the most accurate steering predictions is identified and performance compared to previous neural networks.

KEYWORDS

Autonomous Vehicles, Convolutional Neural Network, Deep Learning, Perception, Self-Driving Cars.


Divide-and-Conquer Federated Learning under Data Heterogeneity

Pravin Chandran, Raghavendra Bhat, Avinash Chakravarthy, Srikanth Chandar, Intel Technology India Pvt. Ltd, Bengaluru, India

ABSTRACT

Federated Learning allows training of data stored in distributed devices without the need for centralizing training-data, thereby maintaining data-privacy. Addressing the ability to handle data heterogeneity (non-identical and independent distribution or non-IID) is a key enabler for the wider deployment of Federated Learning. In this paper, we propose a novel Divide-and-Conquer training methodology that enables the use of the popular FedAvg aggregation algorithm by over-coming the acknowledged FedAvg limitations in non-IID environments. We propose a novel use of Cosine-distance based Weight Divergence metric to determine the exact point where a Deep Learning network can be divided into classagnostic initial layers and class-specific deep layers for performing a Divide and Conquer training. We show that the methodology achieves trained-model accuracy at-par with (and in certain cases exceeding) the numbers achieved by state-of-the-art algorithms like FedProx, FedMA, etc. Also, we show that this methodology leads to compute and bandwidth optimizations under certain documented conditions.

KEYWORDS

Federated Learning, Divide and Conquer, Weight divergence.


Suppression of Power Frequency Noise of Vibration Signal with Windowed Interpolation Method

Zhang Xiaotao, Automobile College, Guangdong Mechanical & Electronical college of Technology, Guangzhou, China

ABSTRACT

Power frequency noise interference is widespread. In order to suppress power frequency noise coupled into vibration signal, windowed interpolation method is introduced. Three typical window functions are selected, including rectangular window, Hanning window and six-term cosine window. The influence of window function on the interpolation results of power frequency noise is simulated and analyzed, being pointed out that different window functions should be selected when the spectrum spacing between power frequency noise and useful components is different. According to the actual vibration signal, the interpolation results of power frequency noise with three window functions are compared, the result of interpolation is the most accurate when Hanning window is added to the signal. Simulation and practice show that the windowed interpolation method is effective for power frequency noise suppression.

KEYWORDS

Power frequency noise, Windowing and interpolating, Vibration signal, Window function.


Efficient Reformatting of Text Spread out in DOM Trees

Changfeng Yu, Cheng Zhang, Hao Zhang and Jie Wang, University of Massachusetts, Lowell, USA

ABSTRACT

We present an efficient algorithm for reformatting text contained in multiple nodes that are spread out on a DOM tree of an HTML file automatically converted from a typeset document. Reformatting text on the fly is needed for certain text-mining applications. A naive approach would traverse the DOM tree multiple times, failing to meet the requirement of real-time reformatting. Our algorithm meets the real-time requirement by indexing text nodes and sentences with a pair of tag holders inserted into each text node to allow fast reformatting.

KEYWORDS

DOM tree, reformatting, indexing.


COVID-19 Twitter Sentiments across the USA in August 2020

Umesh R. Hodeghatta, Ph.D and Sanath Vijay, Northeastern University, Boston, MA, USA; NU-Sigma U2 Analytics Lab, India

ABSTRACT

COVID-19 has drastically affected the entire nation. This study involved collecting tweets and analyzing the COVID tweets for August 2020. The aim was to understand whether people have expressed sentiments related to COVID-19 across all the states of the United States and find any correlation between the sentiment tweets and the number of actual cases reported. Around 400000 COVID-19 Twitter data was collected for August 2020 from the primary Twitter database. A simple NLP-based unigram sentiment analyser, a novel approach different from the traditional machine learning approach, was adopted to identify twitter sentiments. The results indicate that tweets related to COVID demonstrate the two types of sentiments, one related to the deaths and the other about the COVID symptoms. Furthermore, the results show that the sentiments for each category vary from State to State. For example, states of New York, California, Texas are higher tweets sentiments regarding expressing death sentiment, and states of New York, California, Nevada, are higher regarding sentiments of expressing COVID-19 symptoms with an accuracy of 83%. As a part of the research, a new sentiment scorecard was created to provide a sentiment score based on the sentiments of the tweets expressed to the actual reported death cases. The sentiment scores for the ‘symptoms’ class are higher for Maryland, New Jersey, and Oregon, whereas sentiments scores for the 'death' class are higher for Virginia, Delaware, and Hawai. These sentiment scores indicate that the Twitter users of these states are actively tweeting about symptoms and deaths even though the actual reported cases are less in these states. The analysis results also found no or little correlation between the COVID Tweets and the number of COVID death cases reported across all the states.

KEYWORDS

COVID 19, Twitter Behaviour, Twitter Analytics, Sentiment Analysis, Big Data Analytics, Business Analytics.


Impact Of Blockchain Technology In HealthCare sector During COVID-19 Pandemic

Zeba Mahmood, Kaunas Technology University, Lithuania

ABSTRACT

Globally, the pandemic has affected management of risks. Progressively Blockchain is being applicable over the management of healthcare, as an imperative method for improving organizational protocols and for providing the convenient support for a productive and efficient decision-making process hinge on facts. In healthcare, different approaches to emergency preparedness can be recognized; indeed, each emergency is distinguished by different stages. In healthcare, we intend to substantiate blockchain and recommend a trace route on a side of a COVID19-safe clinical proceeding. In alliance along artificial intelligence systems, the adoption of blockchain enables the development of a generalized predictive framework which can be conducive to pandemic risk constraints in the national territory. In future digital healthcare, blockchain may play a strategic role: explicitly, it will be responsible to enhance COVID19-safe clinical proceeding. The primary approaches obtainable from various blockchain-based models, and distinctly those associated by clinical undertaking, have been documented here and critically discussed. We have explored that blockchain can overcome the limitations of the existing system and thereby assist individuals in the future throughout the current COVID-19 pandemic either on the assumption of further more infectious conditions. We believe that in real infectious disease outbreaks, blockchain technology would be capable to perform an outstanding part in the future.

KEYWORDS

COVID–19, Blockchain, pandemic, healthcare.


menu
close