Towards efficient similarity search with semantic hashing techniques

Speaker: Mr Luyao Liu
Host: A/Prof Helen Huang

Seminar Type:  PhD Thesis Review

Abstract: 

The explosive growth of multimedia data scale brings considerable pressure for modern information technology on data management, data storage and computational operation. Facing multitudinous data resources, providing real-time analysis of the data to enhance in-depth understanding of the latent values has become a highly concerned issue and unprecedented challenging tasks in academia and industries around the world. Particularly, similarity search is a fundamental research problem in multimedia retrieval, computer vision, data mining and machine learning. It turns out to be a challenging task when employing similarity search on large-scale multimedia data due to the infeasible computational complexity and memory cost.  

Fortunately, hashing, as an advanced indexing technique, has drawn substantial attention in the past decade due to its promising performance in both efficiency and accuracy while satisfying memory savings. However, several technical issues of existing solutions have not yet been well addressed in many real-world applications, such as large-scale multimedia retrieval and fashion recommendation. In this thesis, it is concentrating on building discriminative hashing models to tackle different challenging tasks of similarity search.

Biography: 

Mr Luyao Liu obtained his B.S in Electronic Information Engineering from Beijing University of Aeronautics & Astronautics in 2010, and the M.S in Engineering Science Management from the University of Queensland in 2013. 

Currently he is a PhD student in DKE group at ITEE of UQ, under the supervision of A/Prof Helen Huang.