Workshop Program

 

International Workshop on SAI-AAAI 24

Workshop Chair:  Xiaoli Li, Joey Tianyi Zhou, Callie Hao, Vijay Janapa Redi, Yung-Hsiang Lu
PC Chair:  Zhenghua Chen (A*STAR)

Session Chair: Fang Yuan (SMU)

Time (Local Time)

Title

Presenter/Author

9:00am - 9:10am

Opening Remarks

Organizers

9:10am - 10:00am

Keynote Presentation: Tensor Networks in the Era of Sustainable AI

Prof. Zenglin Xu

Harbin Institute of Technology

10:00am - 10:20am

Oral 1: She had Cobalt Blue Eyes: Prompt Testing to Create Aligned and Sustainable Language Models

Veronica Chatrath, shaina raza

10:20am - 10:40am

Oral 2:  BoolNet: Towards Energy-Efficient Binary Neural Networks Design and Optimization

Nianhui Guo, Joseph Bethge, Haojin Yang, Kai Zhong, Xuefei Ning, Christoph Meinel, Yu Wang

10:40am - 11:10am

Break & Poster Session

 

11:10am -11:30am

Oral 3:  The Power of Training: How Different Neural Network Setups Influence the Energy Demand

Daniel Geißler, Bo Zhou, Mengxi Liu, Sungho Suh, Paul Lukowicz

11:30am - 11:50am

Oral 4:  Explanations from Large Language Models Make Small Reasoners Better

Shiyang Li, Jianshu Chen, yelong shen, Zhiyu Chen, Xinlu Zhang, Zekun Li, Hong Wang, Jing Qian, Baolin Peng, Yi Mao, Wenhu Chen, Xifeng Yan

11:50am - 12:10pm

Oral 5: CoSS: Co-optimizing Sensor and Sampling Rate for Data-Efficient AI in Human Activity Recognition

Mengxi Liu, Zimin Zhao, Daniel Geißler, Bo Zhou, Sungho Suh, Paul Lukowicz

12:10pm - 12:30pm

Oral 6: TinyM$^2$Net-V3: Memory-Aware Compressed Multimodal Deep Neural Networks for Sustainable Edge Deployment

Hasib-Al Rashid, Tinoosh Mohsenin

12:30pm – 2:00pm

Lunch

 

2:00pm – 2:50pm

Keynote Presentation: Kapacity - An Open Source Solution for Green Data Centers with Predictive AutoScaling

Dr. James Zhang, Managing Director at Ant Group

2:50pm - 3:00pm

Poster 1:  Local Expert Diffusion Models for Efficient Training in Denoising Diffusion Probabilistic Models

Seoungyoon Kang, Yunji Jung, Hyunjung Shim

3:00pm - 3:10pm

Poster 2:  Effective and Sparse Count-Sketch via k-means clustering

Yuhan Wang, Zijian Lei, Liang Lan

3:10pm - 3:20pm

Poster 3:  Causal AI Framework for Unit Selection in Optimizing Electric Vehicle Procurement

Chi Zhang, Ang Li, Scott Allen Mueller, Rumen Iliev

3:20pm - 3:30pm

Poster 4:  MADA: Mask Aware Domain Adaptation for Open-set Semantic Segmentation

Keying Zhang

3:30pm - 3:40pm

Poster 5:   CR-Cross: A Novel Approach for Cross Domain Coral Recognitions with Reject Options

hongyong han, Wei Wang, Gaowei Zhang, Mingjie Li, Yi Wang

3:40pm - 3:45pm

Closing Remarks

Organizers

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

First Keynotes

TITLE:  Tensor Networks in the Era of Sustainable AI

Abstract: As AI models become increasingly complex and data-intensive, the demand for computational resources escalates, raising concerns about the carbon footprint and energy efficiency of these technologies. This talk first introduces the concept and diagrams of tensor networks, then discusses how tensor networks, a mathematical framework originally developed in the context of quantum physics, offer a promising pathway to mitigate these challenges. By enabling the efficient representation and manipulation of high-dimensional data, tensor networks significantly reduce the computational cost and energy consumption of large-scale AI operations. The presentation will cover foundational concepts of tensor networks, including their structure, properties, and how they facilitate sparse data representation and dimensional reduction. It will also highlight recent advancements in integrating tensor networks with deep learning models, demonstrating their effectiveness in tasks such as image recognition, action recognition, natural language processing, with a fraction of the computational resources typically required. A special focus will be directed towards the application of tensor networks in the optimization of large language models, illustrating their substantial impact in this specific area. Finally, it concludes with discussion on the future directions and potential challenges in this emerging field.

 

Biography: Dr. Zenglin Xu earned his Ph.D. in Computer Science and Engineering from the Chinese University of Hong Kong. He holds the position of full Professor at the Harbin Institute of Technology, Shenzhen, and is also affiliated with Peng Cheng Laboratory. Dr. Xu's academic journey has led him through several renowned institutions worldwide. He has been affiliated with Michigan State University, Saarland University's Cluster of Excellence, the Max Planck Institute for Informatics, Purdue University, and the University of Electronic Science and Technology of China. With a primary focus on machine learning, Dr. Xu's research extends into its applications in fields such as computer vision, health informatics, and natural language processing. He has published over 180 peer-reviewed articles in leading journals and conferences within the artificial intelligence domain. His dedication and excellence in the realm of research have been recognized multiple times. Notably, he was honored with the Outstanding Student Paper Mention at AAAI 2015, the Best Student Paper Runner-Up at ACML 2016, and the 2016 Young Researcher Award presented by APNNS. In addition to his research contributions, Dr. Xu plays an active role in the academic community. He serves as an Associate Editor for Neural Networks and consistently contributes as an Area Chair or Senior Program Committee member for conferences including ACL, EMNLP, AAAI, and IJCAI. Currently, he holds the position of Vice President for Education at the International Neural Network Society (INNS).

 

 

 

Second Keynotes

 

TITLE:  Kapacity - An Open Source Solution for Green Data Centers with Predictive AutoScaling

Abstract: The exponential growth of cloud computing has led to a disturbing escalation in carbon emissions from data centers, which now contribute to over 3% of global greenhouse gas emissions. This pressing issue calls for urgent action to mitigate their escalating impact on the environment, as well as the strain they impose on the global climate. As part of the efforts towards Ant Group's goal towards carbon neutrality by 2030 and the general ESG (Environmental, Social and Governance) strategy, we focus on improving resource utilization in order to save electricity usage of data centers.  

 

In this talk, both the engineering and algorithmic backgrounds of Ant Group's autoscaling infrastructure (Kapacity) will be introduced.  Our proposed methodologies, such as Full Scaling Automation (FSA), NeuralReconciler for Hierarchical Time Series and Structured Learning and Task-based Optimization for Time Series Forecasting on Hierarchies (SLOTH), not only forecast the resource demands effectively, but can also dynamically adapt resources to accommodate changing workloads in large-scale cloud computing clusters, enabling the clusters in data centers to maintain their desired CPU utilization target and thus improve energy efficiency. Our approaches achieve significant performance improvement compared to the existing work in real-world datasets. These methods have been deployed on large-scale cloud computing clusters in industrial data centers, and according to the certification of the China Environmental United Certification Center (CEC), a reduction of 947 tons of carbon dioxide, equivalent to a saving of 1538,000 kWh of electricity, was achieved during the Double 11 shopping festival of 2022, marking a critical step for our company’s strategic goal towards carbon neutrality.  

Our proposed systems been running robustly and continuously since then, contributing to the energy efficiency of Ant Group's cloud services.   We have further open-sourced Kapacity, which is used for Ant Group's production data centers, and we remain committed to continuously innovating and developing new solutions to contribute towards the creation of a more sustainable technological infrastructure.

 

Biography: Dr. James Zhang is the Managing Director of AI Forecasting and Strategy Platform of Ant Group. Dr. Zhang obtained his Ph.D. degree from Univ. of Ottawa, Canada in Electrical Engineering, and both his Master’s and Bachelor degrees from Zhejiang Univ., China. Before he joined Ant Financial, Dr. Zhang worked on finance-related AI at Bloomberg, helped setting up the AI branch of Bloomberg Labs and initiating the GPU computation farm of Bloomberg. Dr. Zhang worked in various areas including image processing, natural language processing, high-speed hardware development, optical networks, operations research, biometrics, and financial systems.