NTU Singapore
Title of Talk: Memetic Mission Management and Planning Framework
Abstract: In a broader context, a mission can be defined as a significant task with specific objectives in mind. Typically, such a task involves the execution of a set of operations or procedures. There are many tasks in the real-world, particularly in the non-military domain that can benefit from coordinated and cooperative planning (CCP). To achieve this, a platform for CCP should be scalable across applications or problem domains, at the same time drawing upon reusable modules (APIs and memes) to facilitate rapid prototyping of turnkey solution. In this talk, we discuss the overall framework and illustrate using some practical scenarios.
Stellenbosch University, Department of Industrial Engineering, South Africa
Title of Talk: Particle Swarm Optimization for Large Scale Optimization
Abstract: It is known that standard particle swarm optimization (PSO) algorithms do not scale well to large-scale optimization problems. As the number of dimensions of the search landscape increases, the volume of the search space grows exponentially, and as a result, the performance of standard PSO algorithms deteriorate significantly. While a number of successful adaptations to the PSO have been developed to solve such large-scale optimization problems, these approaches have been developed without first gaining a clear understanding of the reasons why the PSO does not scale well. This presentation will analyze the scalability of standard PSO algorithms with the main goal to identify the reasons for its poor scalability. A number of approaches will be discussed to explore how the curse of dimensionality can be addressed for the PSO.
Jahangirnagar University, Bangladesh
Title of Talk: Prospects and Challenges of Deep Learning for Object Detection and Recognition
Abstract: Nowadays machine learning, in particular, the deep neural network has gained huge popularity due to its outstanding performance in a wide range of applications in computer vision for diverse image-based object detection, recognition, and classification tasks. It bears competitive as well as the conflicting strategy that provides a way to learn deep representations without extensive annotated training data. In this talk, 1) a brief overview of different deep neural networks, such as the convolutional neural network (CNN), recurrent neural network (RNN) and generative adversarial network (GAN) will be presented; 2) Applications, prospects as well as challenges of deep neural network for different object detection, recognition and classification will be highlighted; 3) Promising research directions and unsolved tasks will be discussed to serve as guidelines for future work in solving practical problems.
Cankaya University
Title of Talk: On fractional optimization and some applications
Abstract: The fractional calculus has a fundamental role in better understanding of the dynamics of complicated nonlinear dynamical systems. A short presentation of the properties of the fractional calculus operators will be given. Besides, a novel sign fractional least mean square algorithms are presented for ease in hardware implementation by utilizing sign function to input data and estimation error corresponding to first and fractional-order derivative terms in weight update mechanism of the standard approach. Finally, the comparison of the results from true parameters of the model shows clearly the worth of the scheme in terms of accuracy, convergence and robustness.
Independent Consultant France
Title of Talk: Iterative optimisation: the questionable balance mantra
Abstract: In iterative optimisation a classical belief is that for efficiency a fine balance between local intensive exploitation and global exploration should be achieved. However, to date there is no rigorous approaches (theoretical or experimental) that support it. We present here how exploitation and exploration can be defined and measured in order to really monitor the progress of their ratio. Preliminary results show that it may be far from what one would expect.
Indian Statistical Institute, Kolkata, India
Title of Talk: Deep Generative Adversarial Networks with Application to Class-Imbalanced Learning
King Mongkut's University of Technology Thonburi Bangkok, Thailand
Title of Talk: Innovative Cognitive Computing
Abstract: With rapid advancement in AI, innovative cognitive computing (which we coined “IC2 - I see too”) is needed to develop effective, efficient and safe technologies for the benefits of mankind. Traditionally, cognitive computing is a sub-field of AI, as coined by IBM, and typically involves the use of a computer system, such as IBM Watson supercomputer, to aid in human decision making. However, the current trend is also about extending AI by augmenting it with cognitive abilities of human and possibly other entities. Innovative Cognitive Computing (IC2) integrates the inherently human attribute of cognition with computing which is arguably superior using machines for augmentation. The vision of IC2 is to enable us to “see” the (new) world in a different light. This talk provides an overview of the latest innovative scientific research applied to the multidisciplinary field of cognitive computing.
Indian Institute of Technology Kanpur, India
Title of Talk: Intelligent Health Monitoring of Machines: An Artificial Intelligence Framework
Abstract: Artificial Intelligence has emerged as a key player in the rapidly changing paradigm of the fourth industrial revolution. This rapid transformation of industries can be attributed to smart sensors. Data captured by these sensors can be processed and analyzed by AI to achieve a vast array of objectives ranging from fault diagnosis to residual life prediction. This keynote covers the future development of health monitoring systems based on the principle of advanced AI techniques. AI-enabled systems have shown tremendous success with the introduction of deep learning and neural network. These systems enabled industries to form complex and integrated digital infrastructures with smart decision-making modules. A brief overview of emerging AI techniques, including deep learning, transfer learning, and a few more, will be discussed. This keynote will also share the novel trends of deep learning strategies and their applications, particularly in the field of Machine health monitoring systems. Machine Health monitoring encompasses diagnostics as well as prognostics of a machine and its components. Assessment of a machine on these two parameters is very critical to anticipate failures, minimize economic loss, and subsequently prevent any unexpected accidents. This keynote will also cover how to deal with the challenges of machine health monitoring using deep learning.
Faculty of Engineering & Information Technology
University of Technology Sydney, Australia
Title of Talk: Evolutionary (Big) Data Analytics
Abstract: Evolutionary computation (EC) has been widely used during the last two decades and has remained a highly-researched topic, especially for complex real-world problems. The EC techniques are a subset of artificial intelligence, but they are slightly different from the classical methods in the sense that the intelligence of EC comes from biological systems or nature in general. The efficiency of EC is due to their significant ability to imitate the best features of nature which have evolved by natural selection over millions of years. The central theme of this presentation is about EC techniques and their application to civil structures and infrastructures. On this basis, the presentation I about an evolutionary approach called genetic programming for data mining. Applied evolutionary computing in the data mining field will be presented, and then their new advances will be mentioned such as big data mining. Here, some of my studies on big data mining and modeling using EC and genetic programming, in particular, will be presented. As a case study, EC application in one structural health monitoring problem, inverse identification, will be introduced. And then, the application of EC for response modeling of a new structural system under seismic loads will be explained in detail to demonstrate the applicability of these algorithms on a complex real-world problem.
Department of Electrical Engineering, J.N.V. University, Jodhpur.
Title of Talk: Convolutional Neural Network: The Biologically inspired Deep Neural Network
Abstract: Biological neural systems inspire convolutional Neural Networks (CNN's), like other neural networks. However, these networks show a more considerable resemblance to the biological system. The connectivity patterns of CNN's in Convolutional layers resemble the connectivity pattern of the mammalian visual system. Further, like the mammalian visual system, these networks build the representation of input image hierarchically. Also, these networks are a class of deep neural networks having wide applications in tasks related to image classification, segmentation, registration, and object detection, etc. Although the first computational model based on neurophysiological findings of the mammalian visual system came into existence way back in the year 1980, these networks did not achieve noteworthy success in their use for real applications until the year 2012. The recent success of CNNs in accomplishing the tasks, as mentioned above, is attributed to the advancement of high technology central processing units (CPUs), graphical processing units (GPUs), availability of massive amounts of data, and development of new CNN architectures. Now, they are state-of-art networks specialized in discovering good representations for images, and for doing this, they do not require prior knowledge or human effort. CNN directly processes two or three-dimensional images; this keeps the structural and configural information of the image intact. Currently, CNN's are the most researched machine learning networks in medical image analysis, even though many medical image datasets are small scale datasets. To overcome the limited availability of the medical image datasets, researchers take advantage of their property that the representations learned by them are mostly generic, making them suitable for the transfer learning approach. To achieve this, CNN's are first trained on a large scale well-annotated datasets of a different domain. Then this acquired knowledge is transferred to make CNN's suitable for small datasets available for medical imaging tasks.
This talk aims to introduce the participants with the resemblance of CNN with the biological system, architecture, and working of the primary Convolutional Neural Network. The talk will (i) describe the need to improvise the underlying architecture of CNN's, (ii) throw light on different architectures of CNN, (iii) provide the details on the strategies adopted for regularization in CNN's. Further, it will discuss the concept of transfer learning and its specific application for medical image classification tasks (including the detection of COVID-19).
Senior Editor, Springer, India
Title of Talk: Nuances and Tools of Scientific Publishing
Abstract: The importance of research publishing can be defined by a simple quote of Gerard Piel, which says “Without publication, science is dead.” The first scientific journal was published in 1665 and we have traveled 355 years since then. In the last 20 years, science and reporting of science have undergone revolutionary changes. Computerization and the Internet have changed the traditional ways of reading and writing. Hence, it is very important for scientists and students of the sciences in all disciplines to understand the complete process of writing and publishing scientific papers in good journals. There is also a downside of digital publishing. The principal challenge for publishers is to handle ethical issues and it is of utmost importance for the authors to understand the ethical practices involved in the process. The talk is designed to provide information on different elements of publishing and also on how to make use of various author services for the publishing work.
Type: Error
Message: Object of class CI_Loader could not be converted to string
Filename: /home1/scrsiueb/public_html/cis2020/application/views/keynote-speakers.php
Line Number: 39
Backtrace:
File: /home1/scrsiueb/public_html/cis2020/application/views/main.php
Line: 2
Function: view
File: /home1/scrsiueb/public_html/cis2020/application/core/MY_Controller.php
Line: 55
Function: view
File: /home1/scrsiueb/public_html/cis2020/application/controllers/Keynote_speakers.php
Line: 15
Function: datainfo
File: /home1/scrsiueb/public_html/cis2020/index.php
Line: 315
Function: require_once