NeurIPS 2019 Expo Workshop
Challenges, Advances and Opportunities for Machine Learning Algorithms and Infrastructure at Alibaba
Sponsor: Alibaba Group
Yangqing Jia (Alibaba Group), Rong Jin (Alibaba Group), Wei Lin (Alibaba Group), Yuan Qi (Ant Financial), Le Song (Georgia Institute of Technology & Ant Financial), Shuang Yang (Ant Financial), Hongxia Yang (Alibaba Group)
WORKSHOP DESCRIPTION The last decade has witnessed great success of machine learning in multiple domains, including speech recognition, image classification, video content analysis, search, computational advertisement and finance. Despite the amazing progress, we also run into many challenges when coming to practical applications of machine learning technologies.
In this workshop, we would like to share some of the key developments of machine learning at Alibaba Group that explicitly address limitations of the existing machine learning techniques, both on algorithms and infrastructure. More specifically, we will discuss and share our practical experiences of (a) A Data-driven Approach for GPU Resource Optimization over A Large-scale AI Cluster; (b) FusionStitching: Boosting Execution Efficiency of Memory Intensive Computations for DL Workloads; (c) AliGraph: A Comprehensive Graph Neural Network Platform; (d) Shared Learning: Cross-Organization Joint Machine Learning without Compromising Privacy or Data Security; (e) Machine Learning for All-Inclusive Finance; (f) Extremely Large Scale Image Classification with Long- tailed Noisy Data. These talks almost cover the biggest challenges and most recent advancements in our core departments and we hope that this workshop will nourish further development in these areas.
WORKSHOP ORGANIZERS • Yangqing Jia, AliCloud, Alibaba Group, China • Rong Jin, DAMO Academy, Alibaba Group, USA • Wei Lin, AliCloud, Alibaba Group, China • Yuan Qi, Ant Financial, China • Le Song, Georgia Institute of Technology & Ant Financial • Shuang Yang, Ant Financial, USA • Hongxia Yang, DAMO Academy, Alibaba Group, USA
MORE INFO https://102.alibaba.com/downloadFile.do?file=1567596830365/NIPS_Expo.pdf