Workshop on Split Learning for Distributed Machine Learning (SLDML’21)

March 4-5, 2021 10:00 AM EST onwards (MIT, Virtual)

Join Slack channel on #split_learning for discussion

Workshop Registration Form

Livestream Link

Overview

Friction in data sharing and restrictive resource constraints pose to be a great challenge for large scale machine learning. Recently techniques such as Federated Learning and Split Learning have provided a suite of resource efficient mechanisms for distributed machine learning. The focus of this workshop is to disseminate a lively set of research works around split learning and a sampling of other variants that are of great relevance to this paradigm. These topics are of increasingly large commercial, academic and policy interest.

It is therefore important to build an organic community for this research area, where collaborating researchers share empirical/theoretical insights, code, data, benchmarks and training pipelines in order to advance distributed machine learning. In addition to disseminating recent advances, we plan to identify and document open problems and challenges in split learning and its variants.

Workshop Program

Day 1 - March 4
Time (EST) Title Speakers
10:00 - 10:20 AM Opening Remarks Ramesh Raskar, Praneeth Vepakomma (MIT)
10:20 - 10:50 AM Keynote talk 1 OpenMined
10:50 - 11:05 AM Parallel Training of Deep Networks with Local Updates Michael Laskin, Luke Metz, Seth Nabarrao, Mark Saroufim, Badreddine Noune, Carlo Luschi, Jascha Sohl-Dickstein, Pieter Abbeel, (UC- Berkeley, Google Research, Graphcore Research)
11:05 - 11:20 AM Communication-Efficient Parallel Split Learning Jihong Park, Seungeun Oh, Hyelin Nam, Seong-Lyun Kim, Mehdi Bennis (Deakin University, Yonsei University, University of Oulu)
11:20 - 11:35 AM Coffee Break
11:35 - 11:50 AM Blind Learning: An efficient privacy-preserving approach for distributed learning Gharib Gharibi, Praneeth Vepakomma (TripleBlind, MIT)
11:50 - 12:05 AM Training Neural Networks Using Features Replay Zhouyuan Huo, Bin Gu, Heng Huang (Google, MBZUAI, JD Finance America Corporation, University of Pittsburgh)
12:05 - 1:25 PM Lunch Break
1:25 - 1:55 PM Keynote talk 2 Geeta Chauhan, Facebook AI
1:55 - 2:10 PM Shredder: Learning Noise to Protect Privacy with Partial DNN Inference on the Edge Fatemehsadat Mireshghallah, Mohammadkazem Taram, Prakash Ramrakhyani, Ali Jalali, Dean Tullsen, Hadi Esmaeilzadeh (UC-San Diego, ARM Inc, Amazon)
2:10 - 2:25 PM FedSL: Federated Split Learning on Distributed Sequential Data in Recurrent Neural Networks Ali Abedi and Sheroz S. Khan (KITE, University Health Network, Canada. University of Toronto Canada.)
2:25 - 2:40 PM Differentially Private Supervised Manifold Learning with Applications like Private Image Retrieval Praneeth Vepakomma, Julia Balla, Ramesh Raskar, (MIT)
2:40 - 2:55 PM DISCO: Dynamic and Invariant Sensitive Channel Obfuscation Abhishek Singh, Ayush Chopra, Vivek Sharma, Ethan Z. Garza, Emily Zhang, Praneeth Vepakomma, Ramesh Raskar (MIT, Harvard Medical School)
2:55 - 3:10 PM A Low Complexity Decentralized Neural Net with Centralized Equivalence using Layer-wise Learning Xinyue Liang, Alireza M. Javid, Mikael Skoglund, Saikat Chatterjee, (KTH)
3:10 - 3:25 PM Unleashing the Tiger: Inference Attacks on Split Learning Dario Pasquini, Giuseppe Ateniese, Massimo Bernaschi, (Sapienza Università di Roma, Stevens Institute of Technology)
3:25 - 3:40 PM Interpretable Complex-Valued Neural Networks for Privacy Protection Liyao Xianga , Hao Zhanga , Haotian Mab , Yifan Zhanga , Jie Rena , and Quanshi Zhanga, (Shanghai Jiao Tong University, South China University of Technology)
3:40 - 3:50 PM Break
3:50 - 4:35 PM Hands-on training workshop-Code your SL- Module 1 Abhishek Singh, Praneeth Vepakomma, Ayush Chopra (MIT)
4:35 - 5:20 PM Hands-on training workshop-Code your SL-Module 2 Abhishek Singh, Praneeth Vepakomma, Ayush Chopra (MIT)
Day 2 - March 5
Time (EST) Title Speakers
10:00 - 10:30 AM Keynote talk 3 Supriyo Chakraborty, IBM T.J.Watson Research
10:30 - 10:45 AM Label Leakage and Protection in Two-party Split Learning Oscar Li, Jiankai Sun, Weihao Gao, Hongyi Zhang, Xin Yang, Junyuan Xie, Chong Wang (CMU, ByteDance Inc, University of Washington)
10:45 - 11:00 AM Communication-Efficient Multimodal Split Learning for mmWave Received Power Prediction Yusuke Koda, Jihong Park, Mehdi Bennis, Koji Yamamoto, Takayuki Nishio, Masahiro Morikura (Kyoto Univ, Deakin Univ, Univ. of Oulu)
11:00 - 11:15 AM Distributed Heteromodal Split Learning for Vision Aided mmWave Received Power Prediction Yusuke Koda, Jihong Park, Mehdi Bennis, Koji Yamamoto, Takayuki Nishio, Masahiro Morikura (Kyoto Univ, Deakin Univ, Univ. of Oulu)
11:15 - 11:30 AM Break
11:30 - 11:45 AM Distributed Min–Max Learning Scheme for Neural Networks With Applications to High-Dimensional Classification Krishnan Raghavan, Shweta Garg, Sarangapani Jagannathan and V. A. Samaranayake (Missouri S&T, Argonne National Lab)
11:45 - 12:00 PM Divide and Conquer: Leveraging Intermediate Feature Representations for Quantized Training of Neural Networks Ahmed T. Elthakeb, Prannoy Pilligundla, Alex Cloninger, Hadi Esmaeilzadeh (UC-San Diego)
12:00 - 12:15 PM SplitEasy: A Practical Approach for Training ML models on Mobile Devices Kamalesh Palanisamy (Undergrad), Vivek Khimani (Undergrad), Moin Hussain Moti, Dimitris Chatzopoulos (NIT Trichy, Drexel University, The Hong Kong University of Science and Technology-HKUST)
12:15 - 1:30 PM Lunch Break
1:30 - 2:00 PM Keynote talk 4 Peter Kairouz, Google
2:00 - 2:15 PM Split learning for vertically partitioned data Iker Ceballos, Vivek Sharma, Eduardo Mugica, Abhishek Singh, Praneeth Vepakomma, Ramesh Raskar (Acuratio/MIT)
2:15 - 2:30 PM Split Learning for medical imaging: Multicenter deep learning without sharing patient data Maarten G. Poirot, Praneeth Vepakomma, Ken Chang, Jayashree Kalpathy-Cramer, Rajiv Gupta, Ramesh Raskar (MGH/MIT/Twente/BWH)
2:30 - 2:45 PM TIPRDC: Task-Independent Privacy-Respecting Data Crowdsourcing Framework for Deep Learning with Anonymized Intermediate Representations Ang Li, Jiayi Guo, Huanrui Yang, Yiran Chen (Duke University, Tsinghua University)
2:45 - 3:00 PM Coffee Break
3:00 - 3:15 PM SplitFed: Blending federated learning and split learning Chandra Thapa, MAP Chamikara, and Seyit Camtepe (CSIRO Data61)
3:15 - 3:30 PM Split Learning on FPGAs Hannah Kathleen Whisnant, Praneeth Vepakomma, Ramesh Raskar (United States Military Academy at West Point, IDSS, TPP, MIT)
3:30 - 3:45 PM A preliminary assessment of federated ML for fraud detection. NTTData Italy- AI, IoT & VR-Adriano Manfre, Claudia Lunini, Federico Ungolo, Riccardo Musmeci
3:45 - 4:15 PM Hands-on training workshop-Code your SL-Module 3 Abhishek Singh, Praneeth Vepakomma, Ayush Chopra (MIT)
4:15 - 5:00 PM Closing remarks & Discussion + Social

Important Information

When: March 4-5, 2021
Where: Virtual
Questions: vepakom@mit.edu (email with subject: SL Workshop/Query)
Workshop Organizers: Praneeth Vepakomma, Abhishek Singh, Vitor Pamplona, Otkrist Gupta, Ayush Chopra, Emily Zhang, Yaateh Richardson, Vivek Sharma, Subha Nawer Pushpita, Julia Balla, and Ramesh Raskar