The 5th International Workshop
on Innovative Algorithms for Big Data
IABD2019- The 5th International Workshop on Innovative Algorithms for Big Data
October 30 - November 1 , 2019
The International Workshop on Innovative Algorithms for Big Data is intended to provide an international forum for researchers working in the areas of algorithms, data structure, and modeling for Big Data and applications using them. All researchers developing core technology, including (but not limited to) sublinear-time algorithms, sublinear-size data structure, and sublinear-size modeling for Big Data are welcome to attend the workshop.
- Pankaj Agarwal (Duke University): "Modeling and Analyzing Large High Resolution Terrain Data"
- Christina Boucher (University of Florida): TBA
- ACC Coolen (King's Colledge London): TBA
- Candy Hsu (National Tsing Hua University): "Domain Adaptation and Generalization via Adversarial Learning"
Domain adaptation aims to learn a model on an unlabeled target domain by referring to a fully-labelled source domain; whereas domain generalization refers to multiple source domains and aims to adapt the learnt model to an unseen target domain. This talk will address issues of both problems and will focus on the application of image classification. To deal with the domain adaptation problem, we adopt the idea of domain adversarial training to minimize the source-target domain shift. Furthermore, in order to ensure the classifier is also target discriminative, we include a generative network to guide the classifier so as to push its decision boundaries away from high density area of target domain. On the other hands, to deal with the domain generalization problem, we propose to leverage the generalization capability by explicitly referring to image content and style from different source domains, and impose an additional constraint on the discriminator to further boost the class-discriminative capability.
- Ilan Newman (University of Haifa): "Some recent results on sublinear algorithm for approximating centrality parameters in graphs"
We discuss several "centrality parameters" associated with vertices in undirected large graphs (social networks). E.g., average distance from a vertex), and other. We will make some comparisons between several centrality parameters. Then we show that such parameters cannot be efficiently approximated for bounded degree networks. However, under a distance oracle (rather just neighbourhood oracle), the complexity becomes much more efficient.
- Hidetoshi Nishimori (Tokyo Institute of Technology): "Acceleration of quantum annealing by non-traditional driving of quantum effects"
Quantum annealing is a quantum-mechanical metaheuristic for combinatorial optimization problems. After a general review of the field, I will describe recent developments in several highly non-trivial protocols to accelerate quantum annealing, including non-stoquastic Hamiltonians, inhomogeneous field driving, reverse annealing, and mid-anneal pausing. These methods are expected, and some are shown, to lead to an exponential acceleration of computation compared to the conventional method. I will explain how such is possible and what we can expect in the near future in the field.
- Nicola Prezza (University of Pisa): TBA