The University\nof Macau (UM) has achieved two major breakthroughs in autonomous driving research.\nA study at the university has provided new insights on autonomous driving under\nbad weather and complex road conditions while another study has developed new learning\nmodels that will be specifically applied to autonomous driving in Macao. The research\nresults will be presented at two top annual conferences of artificial\nintelligence.<\/p>
In recent years,\nthe development of deep learning technology has greatly improved machine\nperception and machine\u2019s cognition of the outside world. However, real world\ntasks often encounter problems when the test scene is different from the\ntraining scene. For example, researchers usually obtain abundant training data\nfrom self-driving vehicles on urban roads under standard driving conditions. However,\nif the trained model is directly applied under rare conditions, such as bad\nweather or bad road conditions, there often would be significant accuracy\ndrops. To tackle this problem, a research project between UM and Baidu has proposed\na method named RIFLE to periodically re-initialise the weight of a classifier\nin a learning mode. RIFLE has been proven to provide more meaningful gradient\nback-propagations. The researchers have also verified the effectiveness of\nRIFLE on various real-world transfer learning tasks, including several modern\narchitectures (ResNet, Inception, MobileNet), and basic computer vision tasks\n(Classification, Detection, Segmentation). UM PhD student Li Xingjian is the\nfirst author of this paper. The paper has been accepted for presentation at 2020\nInternational Conference on Machine Learning, a top conference in the field of\nartificial intelligence.<\/p>
The second\nbreakthrough comes from a study on the compression of CNN models in transfer\nlearning. The study proposed a new compression method in which transfer and\ncompression are used alternately. This method can reduce the complexity of the\nmodel and ensure high accuracy at the same time. By compressing part of the layer\nof the CNN model, the complexity can be further reduced. Experimental results\nverify that FLOPs of ResNet-101 can be reduced by 30 per cent on six target\ndata sets, and the accuracy of compressed model remains almost unchanged. As\nFLOPs of the model are reduced by more than 90 per cent, the accuracy is\nmaintained at around 0.70 on multiple data sets. In comparison, models\ngenerated by other methods basically failed to work. The paper has been\naccepted for presentation at the 2020 International Conference on Learning\nRepresentation. Wang Kafeng, a PhD candidate supervised by Prof Xu Chengzhong,\nis the first author of the paper. The study was jointly conducted by UM, the\nShenzhen Institutes of Advanced Technology under the Chinese Academy of Sciences,\nand Baidu.<\/p>
\n\n\n\n\n\n\n\n\n\n\n\n<\/p>
Macao Connected\nAutonomous Driving (MoCAD), led by UM\u2019s State Key Laboratory of Internet of\nThings for Smart City, is funded by the Macao Funding Scheme for Key R&D\nProjects of the Science and Technology Development Fund, Macao SAR (File no. 0015\/2019\/AKP). The project aims to create a swarm\nintelligence based first-class vehicle platform and an experimental base of\nvehicle-road coordination for autonomous driving in the Guangdong-Hong Kong-Macao\nGreater Bay Area. It is a collaborative effort between UM and leading\ninstitutions in mainland China, including Shenzhen Institutes of Advanced\nTechnology under the Chinese Academy of Sciences, the National University of\nDefense Technology, Baidu, and Shenzhen Haylion Technologies. The project is\nheaded by Prof Xu Chengzhong, dean of UM\u2019s Faculty of Science and Technology.<\/p>
<\/p>
Source: Faculty of Science and Technology<\/p>
Media Contact Information:Communications Office, University of Macau<\/p>
<\/p>
<\/p>
<\/p>
<\/p>
<\/p>
Albee Lei Tel\uff1a(853) 88228004Judite Lam Tel\uff1a(853) 88228022Email\uff1aprs.media@um.edu.moUM Website\uff1awww.um.edu.mo<\/p>
<\/p>\n \n ","protected":false},"excerpt":{"rendered":"
The University of Macau (UM) has achieved two major breakthroughs in autonomous driving research. A study at the university has provided new insights on autonomous driving under bad weather and…<\/p>\n","protected":false},"author":1,"featured_media":19243,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[60],"tags":[],"class_list":["post-19242","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-press-release"],"acf":[],"yoast_head":"\n