ICCS 2018 Main Track (MT) Session 3
Time and Date: 13:15 - 14:55 on 12th June 2018
Chair: Panruo Wu
| Hybrid Genetic Algorithm for an On-Demand First Mile Transit System using Electric Vehicles [abstract]
Abstract: First/Last mile gaps are a significant hurdle in large scale adoption of public transit systems. Recently, demand responsive transit systems have emerged as a preferable solution to first/last mile problem. However, existing work requires significant computation time or advance bookings. Hence, we propose a public transit system linking the neighborhoods to a rapid transit node using a fleet of demand responsive electric vehicles, which reacts to passenger demand in real-time. Initially, the system is modeled using an optimal mathematical formulation. Owing to the complexity of the model, we then propose a hybrid genetic algorithm that computes results in real-time with an average accuracy of 98%. Further, results show that the proposed system saves travel time up to 19% compared to the existing transit services.
|Thilina Perera, Alok Prakash, Chathura Nagoda Gamage and Thambipillai Srikanthan
| Comprehensive Learning Gene Expression Programming for Automatic Implicit Equation Discovery [abstract]
Abstract: Automatic Implicit Equation Discovery (AIED), which aims to automatically find implicit equations to fit observed data, is a promising and challenging research topic in data mining and knowledge discovery. Existing methods for AIED are designed based on calculating derivatives, which require high computational cost and will become invalid when the problem encountered contains sparse training data. To tackle these drawbacks, this paper proposes a new mechanism named Comprehensive Learning Fitness Evaluation Mechanism (CL-FEM). The mechanism is capable of learning knowledge from both the given training data and the disturbed data generated by adding stochastic noise to the training data, to measure the validity and fitting error of a given equation model. The proposed CL-FEM is further integrated with the Self-Learning Gene Expression Programming (SL-GEP), forming a Comprehensive Learning Gene Expression Programming (CL-GEP) to solve AIED problems. The proposed CL-GEP is tested on several benchmark problems with different scales and difficulties, and the experiment results have demonstrated that the CL-GEP can offer very promising performance.
|Yongliang Chen and Jinghui Zhong
| Multi-population Genetic Algorithm for Cardinality Constrained Portfolio Selection Problems [abstract]
Abstract: Portfolio Selection (PS) is recognized as one of the most important and challenging problems in financial engineering. The aim of PS is to distribute a given amount of investment fund across a set of assets in such a way that the return is maximised and the risk is minimised. To solve PS more effectively and more effectively, this paper introduces a Multi-population Genetic Algorithm (MPGA) methodology. The proposed MPGA decomposes a large population into multiple populations to explore and exploit the search space simultaneously. These populations evolve independently during the evolutionary learning process. Yet different populations periodically exchange their individuals so promising genetic materials could be shared between different populations. The proposed MPGA method was evaluated on the standard PS benchmark instances. The experimental results show that MPGA can find better investment strategies in comparison with state-of-the-art portfolio selection methods. In addition, the search process of MPGA is more efficient than these existing methods requiring significantly less amount of computation.
|Nasser Sabar, Ayad Turky and Andy Song
| Recognition and Classification of Rotorcraft by Micro-Doppler Signatures using Deep Learning [abstract]
Abstract: Detection and classification of rotorcraft targets are of great significance not only in civil fields but also in defense. However, up to now, it is still difficult for the traditional radar signal processing methods to detect and distinguish rotorcraft targets from various types of moving objects. Moreover, it is even more challeng-ing to classify different types of helicopters. As the development of high-precision radar, classification of moving targets by micro-Doppler features has become a promising research topic in the modern signal processing field. In this paper, we propose to use the deep convolutional neural networks (DCNNs) in rotorcraft detection and helicopter classification based on Doppler radar signals. We apply DCNN directly to raw micro-Doppler spectrograms for rotorcraft de-tection and classification. The proposed DCNNs can learn the features automati-cally from the micro-Doppler signals without introducing any domain back-ground knowledge. Simulated data are used in the experiments. The experimental results show that the proposed DCNNs achieve superior accuracy in rotorcraft detection and superior accuracy in helicopter classification, outperforming the tra-ditional radar signal processing methods.
|Ying Liu and Jinyi Liu
| Data Allocation based on Evolutionary Data Popularity Clustering [abstract]
Abstract: This study is motivated by the high-energy physics experiment ATLAS, one of the four major experiments at the Large Hadron Collider at CERN. ATLAS comprises 130 data centers worldwide with datasets in the Petabyte range. In the processing of data across the grid, transfer delays and subsequent performance loss emerged as an issue. The two major costs are the waiting time until input data is ready and the job computation time. In the ATLAS workflows, the input to computational jobs is based on grouped datasets. The waiting time stems mainly from WAN transfers between data centers when job properties require execution at one data center but the dataset is distributed among multiple data centers. The proposed novel data allocation algorithm redistributes the constituent files of datasets such that the job efficiency is increased in terms of a cost metric. An evolutionary algorithm is proposed that addresses the data allocation problem in a network based on data popularity and clustering. The number of expected job’s file transfers is used as the target metric and it is shown that job waiting times can be decreased by faster input data readiness.
|Ralf Vamosi, Mario Lassnig and Erich Schikuta