Workshop Program
This is a preliminary program. Changes might occur until the day of the workshop.
| May 24th Friday |
|
|---|---|
| 08:30 AM | Openning |
| 08:30 - 9:30 AM | MPP Session 1 |
| 9:30 - 10:00 AM | Coffee break |
| 10:00 - 11:00 AM | MPP Keynote 1 - Sandip Kundu |
| 11:00 AM - 1:30 PM | Lunch |
| 1:30 - 2:30 PM | MPP Session 2 |
| 2:30 - 3:30 PM | MPP Keynote 2 - - Vladimir Alves |
| 3:30 - 4:00 PM | Coffee break |
| 4:00 - 5:00 PM | MPP Session 3 |
| 5:00 PM | Closing |
Title: Can your Machine Learning System be Hacked?
Abstract:
Extensive integration of machine learning (ML) into critical applications like finance or healthcare demands an investigation of security of ML systems against malicious attacks. The various stages of the machine learning process, from data collection to model deployment present multiple avenues of attack that can compromise the integrity of a system. In this talk, we examine the three fundamental pillars of information security, namely, confidentiality, integrity and availability, with the machine learning process and categorize the various attacks against the three pillars. Confidentiality of an ML system involves securing it against exposure of the model parameters via the observation of responses and also, securing the original training dataset. Ensuring integrity of machine learning systems is a difficult task, but is crucial to prevent exploitation by adversaries. We conclude the talk with a discussion of potential defenses and a brief examination of future concerns.
Short bio - Sandip Kundu:
Sandip Kundu is a Program Director at the National Science Foundation in the CNS division within the CISE directorate. He is serving in this position on leave from the University of Massachusetts at Amherst, where he is a professor in Electrical and Computer Engineering Department. Kundu began his career at IBM Research as a Research Staff Member; then worked at Intel Corporation as a Principal Engineer before joining UMass Amherst as a professor in 2005. He has published over 250 research papers in VLSI design and test, holds several key patents including ultra-drowsy sleep mode in processors, and has given more than a dozen tutorials at various conferences. He is a Fellow of the IEEE, Fellow of the Japan Society for Promotion of Science (JSPS), Senior International Scientist of the Chinese Academy of Sciences and was a Distinguished Visitor of the IEEE Computer Society. He has served as an Associate Editor of the IEEE Transactions on Computers, IEEE Transactions on Dependable and Secure Computing, IEEE Transactions on VLSI Systems and ACM Transactions on Design Automation of Electronic Systems. He has been Technical Program Chair/General Chair of multiple conferences including ICCD, ATS, ISVLSI, DFTS and VLSI Design Conference.
Abstract:
Extensive integration of machine learning (ML) into critical applications like finance or healthcare demands an investigation of security of ML systems against malicious attacks. The various stages of the machine learning process, from data collection to model deployment present multiple avenues of attack that can compromise the integrity of a system. In this talk, we examine the three fundamental pillars of information security, namely, confidentiality, integrity and availability, with the machine learning process and categorize the various attacks against the three pillars. Confidentiality of an ML system involves securing it against exposure of the model parameters via the observation of responses and also, securing the original training dataset. Ensuring integrity of machine learning systems is a difficult task, but is crucial to prevent exploitation by adversaries. We conclude the talk with a discussion of potential defenses and a brief examination of future concerns.
Short bio - Sandip Kundu:
Sandip Kundu is a Program Director at the National Science Foundation in the CNS division within the CISE directorate. He is serving in this position on leave from the University of Massachusetts at Amherst, where he is a professor in Electrical and Computer Engineering Department. Kundu began his career at IBM Research as a Research Staff Member; then worked at Intel Corporation as a Principal Engineer before joining UMass Amherst as a professor in 2005. He has published over 250 research papers in VLSI design and test, holds several key patents including ultra-drowsy sleep mode in processors, and has given more than a dozen tutorials at various conferences. He is a Fellow of the IEEE, Fellow of the Japan Society for Promotion of Science (JSPS), Senior International Scientist of the Chinese Academy of Sciences and was a Distinguished Visitor of the IEEE Computer Society. He has served as an Associate Editor of the IEEE Transactions on Computers, IEEE Transactions on Dependable and Secure Computing, IEEE Transactions on VLSI Systems and ACM Transactions on Design Automation of Electronic Systems. He has been Technical Program Chair/General Chair of multiple conferences including ICCD, ATS, ISVLSI, DFTS and VLSI Design Conference.
Title: What is Computational Storage and what are its applications?
Abstract:
In the age of Big Data, the huge volume of data produced every day is posing a significant challenge of extracting useful information and knowledge from raw data in an efficient way. Past studies have revealed that only 0.5% of all data is ever analyzed due to a lack of effective methods to manage the data. The problem is especially prominent (but not only limited to) in the handling of visual data such as images and videos and storage devices at the center of this transformation. In this talk we introduce computational storage devices. These unique devices provide conventional storage functionality combined with dedicated computing resources and a complete software stack for near-data execution of user applications that avoid costly data movement. By moving data processing tasks closer to where the data resides, computational storage dramatically reduces the storage bandwidth bottleneck, data movement cost, and improve the overall energy efficiency creating an ideal platform for data intensive applications. We will examine different approaches currently being investigated in the academia and industry and provide an overview of the standardization efforts underway. We will then cover multiple use cases from image similarity search to machine learning and Hadoop and examine performance gains, scalability and energy efficiency.
Shot Bio - Vladimir Alves:
Dr. Alves is a co-founder and CTO at NGD Systems where he has been responsible for developing ground-breaking storage and computing technology. Prior to that he was Sr. Director of SSD SoC Development at Western Digital from 2011 to 2014, responsible for all enterprise SSD controller design, verification and validation. Prior to joining WD, Vladimir served as Sr. Director of SSD SoC Development at STEC, and was instrumental in developing the technology that put STEC in the leadership position as an enterprise SSD solutions provider. From 2000 to 2005 Dr. Alves was a Scientist and Director of Engineering at MorphoTechnologies a start-up company in California that developed an innovative architecture for a highly configurable and massively parallel DSP array used in high performance signal processing for base stations.
Vladimir obtained his Ph.D. In Microeletronics in 1992 from the National Polytechnic Institute in Grenoble, France. He subsequently held the positions of Associate Professor at the Grenoble Institute of Technology-ENSERG (Grenoble, France), Eurochip Lecturer at the University of Aveiro, Portugal and Professor at the Federal University of Rio de Janeiro. He is the author of more than 30 scientific publications and holds several patents in US and Europe.
Abstract:
In the age of Big Data, the huge volume of data produced every day is posing a significant challenge of extracting useful information and knowledge from raw data in an efficient way. Past studies have revealed that only 0.5% of all data is ever analyzed due to a lack of effective methods to manage the data. The problem is especially prominent (but not only limited to) in the handling of visual data such as images and videos and storage devices at the center of this transformation. In this talk we introduce computational storage devices. These unique devices provide conventional storage functionality combined with dedicated computing resources and a complete software stack for near-data execution of user applications that avoid costly data movement. By moving data processing tasks closer to where the data resides, computational storage dramatically reduces the storage bandwidth bottleneck, data movement cost, and improve the overall energy efficiency creating an ideal platform for data intensive applications. We will examine different approaches currently being investigated in the academia and industry and provide an overview of the standardization efforts underway. We will then cover multiple use cases from image similarity search to machine learning and Hadoop and examine performance gains, scalability and energy efficiency.
Shot Bio - Vladimir Alves:
Dr. Alves is a co-founder and CTO at NGD Systems where he has been responsible for developing ground-breaking storage and computing technology. Prior to that he was Sr. Director of SSD SoC Development at Western Digital from 2011 to 2014, responsible for all enterprise SSD controller design, verification and validation. Prior to joining WD, Vladimir served as Sr. Director of SSD SoC Development at STEC, and was instrumental in developing the technology that put STEC in the leadership position as an enterprise SSD solutions provider. From 2000 to 2005 Dr. Alves was a Scientist and Director of Engineering at MorphoTechnologies a start-up company in California that developed an innovative architecture for a highly configurable and massively parallel DSP array used in high performance signal processing for base stations.
Vladimir obtained his Ph.D. In Microeletronics in 1992 from the National Polytechnic Institute in Grenoble, France. He subsequently held the positions of Associate Professor at the Grenoble Institute of Technology-ENSERG (Grenoble, France), Eurochip Lecturer at the University of Aveiro, Portugal and Professor at the Federal University of Rio de Janeiro. He is the author of more than 30 scientific publications and holds several patents in US and Europe.
Exploring the Equivalence between Dynamic Dataflow Model and Gamma - General Abstract Model for Multiset mAnipulation
Rui Rodrigues de Mello Junior (Federal University of Rio de Janeiro, Brazil)
Leandro Santiago Araújo (Federal University of Rio de Janeiro, Brazil)
Tiago Assumpção de Oliveira Alves (State University of Rio de Janeiro, Brazil)
Leandro Augusto Justen Marzulo (State University of Rio de Janeiro, Brazil)
Gabriel Antoine Louis Paillard (Federal University of Paraná, Brazil)
Felipe Maia Galvão França (Federal University of Rio de Janeiro, Brazil)
A Reinforcement Learning Scheduling Strategy for Parallel Cloud-based Scientific Workflows
André Nascimento (Fluminense Federal University, Brazil)
Victor Olimpio (Fluminense Federal University, Brazil)
Vítor Silva (Federal University of Rio de Janeiro, Brazil)
Aline Paes (Fluminense Federal University, Brazil)
Daniel de Oliveira
Rui Rodrigues de Mello Junior (Federal University of Rio de Janeiro, Brazil)
Leandro Santiago Araújo (Federal University of Rio de Janeiro, Brazil)
Tiago Assumpção de Oliveira Alves (State University of Rio de Janeiro, Brazil)
Leandro Augusto Justen Marzulo (State University of Rio de Janeiro, Brazil)
Gabriel Antoine Louis Paillard (Federal University of Paraná, Brazil)
Felipe Maia Galvão França (Federal University of Rio de Janeiro, Brazil)
A Reinforcement Learning Scheduling Strategy for Parallel Cloud-based Scientific Workflows
André Nascimento (Fluminense Federal University, Brazil)
Victor Olimpio (Fluminense Federal University, Brazil)
Vítor Silva (Federal University of Rio de Janeiro, Brazil)
Aline Paes (Fluminense Federal University, Brazil)
Daniel de Oliveira
A Reconfigurable Ray-Triangle Vector Accelerator for Emerging Fog Architectures
Adrianno Sampaio (State University of Rio de Janeiro, Brazil)
Alexandre Sena (State University of Rio de Janeiro, Brazil)
Alexandre Nery (University of Brasília, Brazil)
Stream Processing on Multi-Cores with GPUs: Parallel Programming Models' Challenges
Dinei Rockenbach (Três de Maio Faculty and Pontificial Catholic University of Rio Grande do Sul, Brazil)
Charles Stein (Três de Maio Faculty, Brazil)
Dalvan Griebler (Três de Maio Faculty and Pontificial Catholic University of Rio Grande do Sul, Brazil)
Gabriele Mencagli (University of Pisa, Italy)
Massimo Torquati (University of Pisa, Italy)
Marco Danelutto (University of Pisa, Italy)
Luiz Fernandes (Pontificial Catholic University of Rio Grande do Sul, Brazil)
Adrianno Sampaio (State University of Rio de Janeiro, Brazil)
Alexandre Sena (State University of Rio de Janeiro, Brazil)
Alexandre Nery (University of Brasília, Brazil)
Stream Processing on Multi-Cores with GPUs: Parallel Programming Models' Challenges
Dinei Rockenbach (Três de Maio Faculty and Pontificial Catholic University of Rio Grande do Sul, Brazil)
Charles Stein (Três de Maio Faculty, Brazil)
Dalvan Griebler (Três de Maio Faculty and Pontificial Catholic University of Rio Grande do Sul, Brazil)
Gabriele Mencagli (University of Pisa, Italy)
Massimo Torquati (University of Pisa, Italy)
Marco Danelutto (University of Pisa, Italy)
Luiz Fernandes (Pontificial Catholic University of Rio Grande do Sul, Brazil)
Neural Network Frameworks. Comparison on Public Transportation Prediction
Cristina Heghedus (University of Stavanger, Norway)
Antorweep Chakravorty (University of Stavanger, Norway)
Chunming Rong (University of Stavanger, Norway)
Instrumental Data Management and Scientific Workflow Execution: the CEA case study
Francieli Zanon Boito (Univ. Grenoble Alpes, France)
Jean-François Méhaut (Univ. Grenoble Alpes, France)
Thierry Deutsch (Laboratorie de Simulation Atomistique, France)
Brice Videau (Laboratorie de Simulation Atomistique, France)
Frédéric Desprez (Univ. Grenoble Alpes, France)
Cristina Heghedus (University of Stavanger, Norway)
Antorweep Chakravorty (University of Stavanger, Norway)
Chunming Rong (University of Stavanger, Norway)
Instrumental Data Management and Scientific Workflow Execution: the CEA case study
Francieli Zanon Boito (Univ. Grenoble Alpes, France)
Jean-François Méhaut (Univ. Grenoble Alpes, France)
Thierry Deutsch (Laboratorie de Simulation Atomistique, France)
Brice Videau (Laboratorie de Simulation Atomistique, France)
Frédéric Desprez (Univ. Grenoble Alpes, France)