Physically Unlonable Functions (PUFs) have been intensively examined as a novel solution to these challenges of the traditional cryptographic protocol implementation. PUF is a physical entity embedded in a hardware device. A PUF instance has the following special property: the generation of the corresponding response (the output) of a given challenge (the input) completely depends on the intrinsic process variation effects of the hardware device where it is embedded, and the process variation induced features are static random and non-controllable. As a result, ideally, the challenge-response behaviors of different PUF instances are distinct, unpredictable, and it is infeasible to clone another PUF instance with identical challenge-response behavior. In this tutorial, we would explore design challenges, operating principles, threats on PUF circuits. More specifically, the tutorial would cover the following topics:Fundamentals of PUF, Security Analysis of PUFs and Machine learning based attacks on PUFs.

The tutorial starts with the explanation on intrinsic process variations of physical material and exploiting it to use like "device fingerprint" on devices. The different PUF characteristics include unclonability, reproducibility, and unpredictability. The quality metrics of PUF e.g. Uniformity, Uniqueness, and Reliability will also be discussed. The tutorial would focus on the security evaluation of APUF, XOR-APUF, LSPUF, MPUF and IPUF with respect to cryptanalysis based attacks and machine learning based modeling attacks. Such analyses help both the designers and system developers to build a PUF-based secure system.

Finally, the usage of a toolbox called PUFmeter, developed by the University of Florida, will be shown to test the security properties of PUF. This tutorial would help the attendee to understand different security aspects of PUFs, and this understanding ultimately leads to the design of secure PUF circuits.

This tutorial discusses an interesting conjuncture of research in the domain of Internet of Things (IoT) and Machine Learning (ML). IoT combines the technological advancements in the domain of embedded systems, sensors and actuators, wireless communication and networking. The tutorial aims to bring together the research community working in the domain of Internet of Things (IoT) and Machine Learning (ML) for developing futuristic smart applications. The following are the learning objectives of the tutorial:

Beginners : the tutorial will impart basic knowledge about Internet of Things (IoT), Machine Learning (ML), common terminologies, and the use cases enabled by integrating these two technologies.

Practitioners it will provide a solid foundation to compare and evaluate existing approaches for building IoT-ML applications. Developers it will prescribe various tools and technologies for IoT-ML applications development.

Researchers it will provide a discussion forum concerning all the aspects of designing and developing an ecosystem for IoT-ML applications.

The tutorial should be of interest to both the professionals and academic community working in the domain of IoT and Machine Learning. The partici- pants with all levels of experience / knowledge in the domain of IoT and ML are encouraged to attend. The basic knowledge of IoT and Machine Learning is suf- ficient to attend the tutorial.

In this Malware Analysis tutorial, the participants will be learning about static and dynamic malware analysis and tools and more general subjects such as what kinds of malware are around and how antivirus software works. This tutorial will try to answer the following question.

1. 1. Why perform Malware Analysis?
2. 2. How does antivirus software work?
3. 3. How to do basic Malware Analysis?
4. 4. What are Static Malware Analysis tools available?
5. 5. What are Dynamic Malware Analysis tools available?

Participants will be taught to use existing tools and techniques and also some recent researches will also be discussed in the tutorial.
Outline

1. 1. Introduction of Malware.
2. 2. A short history of Malware (Virus to Malware).
3. 3. 1st Generation Malware.
4. 4. 2nd Generation Malware.
5. 5. Traditional defense system.
6. 6. Basic Malware Analysis.
7. 7. Static Malware Analysis.
8. 8. A case study (CodeRed) of static Malware Analysis.
9. 9. Dynamic Malware Analysis.
10. 10. A case study (WannaCry) of dynamic Malware Analysis.
11. 11. Machine Learning in Malware Identification and Classification.
12. 12. Recent research to defend Advanced Malware.

Requirement: Researcher, student, academician or corporate employee, having basic knowledge of Operating system (WINDOWS) and understanding of Assembly codes & C programming language.

First Half(1.5hrs)

Introduction/Refresher to Deep Learning

1. 1. How does Deep Learning relate to AI and Machine Learning?
2. 2. What problems could be solved using Deep Learning?
3. 3. What are the components of a Deep Learning solution?

Designing and Visualizing Deep Networks

1. 1. Using Deep Network Designer to visualize a network.
2. 2. Creating a new network and Analyzing its properties.
3. 3. Importing pre-trained networks from other frameworks.

Second Half(1.5hrs)

Challenges in Deep Learning, and Analysis of Training

1. 1. How do we handle data while training Deep Networks?
2. 2. What to do when training accuracy isn’t as good as expected?
3. 3. Tuning Hyper-parameters and managing experiments.

Deployment to GPUs

1. 1. One representation – multiple platforms for deployment.
2. 2. Workflows for generating GPU compliant code from MATLAB

Q&A

Target Audience: Faculties, Researchers, and Students interested in Deep Learning and its applications

Pre-requisites for Attendees:

1. 1. Attendees need to get their laptop with recent (18b/19a) version of MATLAB installed or MATLAB trial download.
2. 2. No prior knowledge of Deep Learning or Machine Learning is expected.
3. 3. Basic exposure to MATLAB is a plus, but not necessary.
4. 4. It would be useful to have some problems in mind for which you think Deep Learning is a good tool to help solve.

Outcome: After attending this workshop, attendees should be able to:

1. 1. Have a general sense of where Deep Learning fits into AI
2. 2. Know what tools are available to Visualize, Design, and Analyze Deep Networks.
3. 3. Have more confidence in training networks and handling data.

Amith is an Education Technical Evangelist, and he has been with the MathWorks for the past 6 years. He has taken on a variety of roles including technical support, software developer, and most recently, as Technical Evangelist. He has a bachelor’s degree in Electrical Engineering from NITK Surathkal, a Master’s degree in Electrical Engineering from the University of Minnesota, and he’s currently working towards a Master’s degree in Computer Science from Georgia Tech. His areas of interests include computer vision, deep learning, and software engineering.

First Half(1.5hrs)

Graphical System Design Platform: Why, What, How?

Second Half(1.5hrs)

NI GSD platform for Circuits, Controls & Communication Engineering.

Target Audience: Faculties, Researchers and Students interested in preparing for 21st Century System Design Challenges from ECE, EEE, Mech, Civil, EIE, Aero, Biomedical Engineering.

Outcome:: Successful completion of attending the tutorial session, participants will be enabled with NI Software platform access for six months free of cost to build solutions for real world applications.