Keynote Speakers

MOD 2017 Keynote Speakers

  • Assimilated Learning: A Framework for Co-analysis of  Big Data and Smart Data“, Yi-Ke Guo
    Department of Computing, Faculty of Engineering, Imperial College London, UK. Founding Director of Data Science Institute

    The importance of combined analysis of big and smart data has been well recognized and ample research has been conducted with the focus on “data integration” or “data fusion”. However, the aforementioned imbalance in size, context and richness in semantics made the integration at the data level a hard and unsustainable technology. Although there is some remarkable progresses made in studying the interaction of big and smart data and exploring the advantage of both for the mutual enhancement for their analysis, we still lack a systematic study and uniform approach for the joint analysis of both data types. In this talk,  we are introducing Assimilated Learning where smart data and big data will be co-collected and co-analysed in a bi-directionally guided way.

  • “Quantification of Network Dissimilarities and its Practical Implications”, Panos PardalosDepartment of Systems Engineering, University of Florida, USA.  Director of the Center for Applied Optimization.

    In this lecture, we analyze a novel measure that quantifies network dissimilarities by comparing its performance with other well-known tools. The efficacy of this measure, based on Information Theory, depends on the use of rich information extracted from the graphs. We show here that the measure has promising implications in several research areas that include, bioinformatics, climate dynamics, percolation in networks, network robustness and model selection. We perform extensive computational experiments on real and artificial networks. Future research directions, which include applications to multiplex settings, will also be discussed.
     This is joint work with T. Schieber, M.G. Ravetti, and L. Carpi
  • “Recent Advances in Deep Learning”,  Ruslan Salakhutdinov, Machine Learning Department, School of Computer Science at Carnegie Mellon University, USA. Director of AI Research at Apple.

    In this talk I will first introduce a broad class of deep learning models and show that they can learn useful hierarchical representations from large volumes of high-dimensional data with applications in information retrieval, object recognition, and speech perception. I will next introduce models that can find semantically meaningful representations of words, learn to read documents and answer questions about their content. Specifically, I will focus on recurrent neural network models that integrate multi-hop architectures with novel attention mechanism, along with its extensions that make use of external linguistic knowledge. I will further introduce the notion of “Memory” as being a crucial part of an intelligent agent’s ability to plan and reason in partially observable environments and demonstrate a deep reinforcement learning agent that can learn to store arbitrary information about the environment over long time lags. I will show that on several tasks these models significantly improve upon many of the existing techniques.
  • “Socialize Strategies for Bots: when incomplete topology meets efficiency”My ThaiDepartment of Computer & Information Science & Engineering, University of Florida, USA.

    With a huge amount of personal information ripe for the taking in modern Online Social Networks (OSNs), privacy breaches have become a central concern, especially with an introduction of automated attacks by socialbots, which can automatically extract victims’ private content by exploiting social behavior to befriend them.      In this talk, we explore the social strategies of socialbots and see how they can harvest the most of private information using at most k friend requests, modeled as Max-Crawling.  The two main challenges of this problem are how to cope with incomplete knowledge of network topology and how to model users’ responses to friend requests. Accordingly, we present an adaptive approximation algorithm using adaptive stochastic optimization. The key feature of our solution lies in the adaptive method, where partial network topology is revealed after each successful friend request. Thus the decision of whom to send a friend request to next is made with the outcomes of past decisions taken into account. Traditional tools break down when attempting to place a bound on the performance of this technique with realistic user models as it is no longer submodular. Therefore, we additionally introduce a novel curvature-based technique to construct an approximation ratio of for a model of user behavior learned from empirical measurements on Facebook.

  • “Optimization and Management in Manufacturing Engineering”,  Jun Pei, Hefei University of Technology, China.

    With the continuous development of network technology and global economic integration, the competition in manufacturing becomes more and more fierce. There is increasing awareness of the supply chain participants that they have to reinforce the cooperation between each other to improve the competitiveness of supply chain so as to decrease each operation cost. The development of Internet of Things technology provides an information basis of the cooperation between the participants of supply chain. It can not only return the production information to the management center, but also share the information to other participants. The Internet of Things technology pushes the cooperation between supply chain participants to a new level that by using the information effectively can decrease the production cost, increase the profit, improve the satisfaction of customers, and in the end enhance the competitiveness of the whole supply chain. Besides, introducing the technology of the Internet of Things also broadens the theoretical area of the research on scheduling problems. Therefore, how to transform the information value into economic and social value, and use the information acquired by the Internet of Things to obtain efficient production plans becomes the key issues. Based on the background of Aluminum production manufacturing chain in China, we focus on the issues of Optimization and Management in Manufacturing Engineering.

  • “The Cloud and Cognitive Industrial Revolution”
    Vincenzo Sciacca
    Cloud and Cognitive Division – IBM Rome, Italy

    The ability to ask simple questions in natural language, and receive answers based on a huge array of knowledge, intelligently ordered, is opening up a wide range of industries to the benefits of cognitive computing. What is the difference between AI and cognitive? Why Cloud is so important in this new industrial revolution? How humans are involved in the loop? I will talk about cloud and cognitive: a new, emerging industrial trend that leverages on data science, AI technologies and cloud computing.