±¨ ¸æ ÈË£ºEdwin R. Hancock½ÌÊÚ

µ¥Ôª£ºÓ¢¹úÔ¼¿Ë´óѧÍÆËã»ú¿ÆѧѧԺ

»ã±¨¹¦·ò£º2019Äê7ÔÂ8ÈÕ£¨ÖÜÒ»£©10£º00¡«12:00

»ã±¨µØÖ·£ºÐ£±¾²¿ÀÖºõÐÂ¥ѧ˼Ìü

Ô¼ÇëÈË£ºÍõ½¡¼Î

»ã±¨ÌáÒª£º

Techniques based on the classical random walk on a graph, have proved to be extremely powerful in the domain of machine learning for developing algorithms for the analysis of high dimensional data. Examples include data embedding, data clustering and feature extraction. However, quantum walks exhibit properties not shared by their classical counterparts. So whereas the classical walk is both stationary and ergodic, the quantum walk is not. Moreover, the quantum walk admits the possibility of both entanglement and interference. These two attributes of the quantum walk allow us to develop new machine learning algorithms, with very different characteristics to their classical counterparts. For instance, interference allows the symmetry structure of graphs or data represented by graphs to captured in a natural and efficient way. In this talk I will provide a tutorial overview of the quantum walk and its properties, and then outline some of its potential uses in deep learning and complex network analysis.

»ã±¨È˼ò½é:

Edwin R. Hancock½ÌÊÚ£¬ÈÎÖ°ÓÚÓ¢¹úÔ¼¿Ë´óѧ£¬ÊÇÊÀ½çÍÆËã»úÊÓ¾õÓëģʽ¼ø±ðÁìÓòµÄ³ÛÃûר¼Ò£¬¹ú¼Êģʽ¼ø±ðЭ»á£¨International Association for Pattern Recognition, IAPR£©¸±Ö÷ϯ£¬IEEE Fellow£¬IAPR Fellow£¬IET Fellow£¬Fellow of Institute of Physics£¬Í¬Ê±Êǹú¼Êģʽ¼ø±ðÁìÓòȨÍþÆÚ¿¯Pattern RecognitionµÄÖ÷±à¡£ÔøÈÎIEEE Transactions on Pattern Analysis and Machine Intelligence£¬Computer Vision and Image Understanding£¬Image and Vision Computing£¬the International Journal of Complex NetworksµÈ¹ú¼ÊÆÚ¿¯±àί»áίԱ£¬BMVC1994´ó»áÖ÷ϯ£¬BMVC2016·¨Ê½Ö÷ϯ£¬ECCV2006£¬CVPR2008£¬CVPR2014£¬ICPR2004£¬ICPR2016ÁìÓòÖ÷ϯ¡£