IAP 2017 Activities by Category - A.I. and Robotics

Manasi Vartak, PhD Student at CSAIL, Maggie Makar, Grad Student CSAIL

Enrollment: Unlimited: Advance sign-up required
Sign-up by 01/10
Attendance: Participants must attend all sessions
Prereq: Basic ML, Python, basic calculus

Machine learning (ML) is clearly the coolest kid on the block right now, and everyone wants to be friends with ML! However, ML as a field has so many areas and sub-areas, and so much jargon that it is hard for a beginner (or even a grad student in ML) to place all of the problems and techniques in context.

This 4-part class will provide brief overviews of diverse ML areas and discussions comparing and contrasting techniques. Each session will consist of MIT grad students giving 1/2 hr talks on particular topics and a discussion putting those topics in context.

Tentative topics per session are:

- Overview of supervised and unsupervised learning

- Inference

- Bayesian Methods

- Neural Nets

 

Class is geared towards advanced undergraduate and graduate students. It assumes a basic familiarity with ML.

Please sign up here by 1/10: https://goo.gl/forms/ZGAaq3qKtrqATEJC3

 

Sponsor(s): Electrical Engineering and Computer Science, Student Information Processing Board
Contact: Manasi Vartak, 32-G904B, MVARTAK@MIT.EDU

---

Session 1

Jan/24	Tue	03:00PM-05:00PM	36-156
Jan/25	Wed	03:00PM-05:00PM	36-156
Jan/26	Thu	03:00PM-05:00PM	36-156
Jan/27	Fri	03:00PM-05:00PM	36-156

Manasi Vartak - PhD Student at CSAIL, Maggie Makar - Grad Student CSAIL

Andrew Kortina, Co-Founder, Fin and Venmo

Enrollment: Unlimited: Advance sign-up required
Sign-up by 12/29
Attendance: Participants must attend all sessions

Civilizations from the ancient Greeks to our own have looked at work as one of the primary sources of meaning and dignity in life. Perhaps it has been useful evolutionarily to esteem work over hedonism, but if we imagine a world of super-technology, where there is no need for most humans to work to provide for the survival of the species, how will our concepts of human dignity shift?

In this class, we'll consider various perspectives on this question, and the class will culminate in an essay, story, or video that explores the future of dignity in a (perhaps dystopian) world of abundance.

Introduction:

Video Essay: Humans Need Not Apply

Essay: Free Will, Technodeterminism, and Panache 

Essay: "They Say the #1 Killer of Old People is Retirement."

Classical Perspectives:

Excerpts from Homer's Iliad and Odyssey

Plato's Apology

Work, Art, Play:

Film: Jiro Dreams of Sushi 

The Tibetan Sand Mandala

Camus: The Myth of Sisyphus

Alan Watts, Play and Dance

Full syllabus, prep questions, and signup here:

http://kortina.net/ai-mass-automation-and-the-evolution-of-human-dignity/

Sponsor(s): Electrical Engineering and Computer Science
Contact: Andrew Kortina, andrew.kortina@gmail.com

---

Jan/12	Thu	12:00PM-02:00PM	4-153
Jan/13	Fri	12:00PM-02:00PM	4-153

Andrew Kortina - Co-Founder, Fin and Venmo

Max Shen, PhD Student, Alvin Shi, PhD Student, Carles Boix, PhD Student

Jan/10	Tue	05:00PM-06:30PM	4-237
Jan/12	Thu	05:00PM-06:30PM	4-237
Jan/17	Tue	05:00PM-06:30PM	4-237
Jan/19	Thu	05:00PM-06:30PM	4-237
Jan/24	Tue	05:00PM-06:30PM	4-237
Jan/26	Thu	05:00PM-06:30PM	4-237

Enrollment: Unlimited: Advance sign-up required
Sign-up by 01/08
Attendance: Participants welcome at individual sessions

Recent innovations in computational methods for Bayesian inference, captured in probabilistic programming languages such as Stan and Edward, have made the power of fully Bayesian inference accessible beyond expert statisticians. These methods particularly shine in machine learning settings with small-to-medium size datasets and complex prior/domain knowledge.

This class aims to provide a hands-on introduction to applying probabilistic programming to real-world problems. The ideas behind probabilistic programming will be covered, including automatic differentiation, variational inference, Markov Chain Monte Carlo (MCMC), and other inference methods. The engineering of models will also be emphasized through exercises on debugging, model specification, reparameterization, addressing identifiability issues, and model efficiency.

Coding exercises and sample datasets will be provided. Students are also encouraged to bring in their own datasets. All course details are subject to change.

*Prior experience with Python or R recommended, as well as some experience with statistics. The class is geared towards interested undergraduates and graduate students.

*In addition, the first annual Stan Convention is occurring on January 21st at Columbia University ($50 student registration) and some of us will be attending.

*Please register here: https://goo.gl/forms/6Ovz4ferwITj7ak13

*Course Material here: https://github.com/maxwshen/iap-appbml

Contact: Max Shen, MAXWSHEN@MIT.EDU

Dazza Greenwood, JD, Visiting Scientist, MIT Media Lab

Enrollment: By permission of instructors
Sign-up by 01/13
Limited to 40 participants
Attendance: Participants welcome at individual sessions
Prereq: N/A

This course provides a conceptual overview and hands-on projects for understanding and solving legal use cases with data analytics, blockchain and other cryptosystems and a special module on virtual reality for data vizualization. The course includes seminar style lecture/discussion sessions and hands-on, experiential learning through team projects. The course covers:

Legal Analytics, including 1) AI/Machine Learning for solving legal use cases; and 2) Using VR for data-driven visualization of complex financial relationships and legal contexts

Digital Assets, including: 1) Ownership rights, valuation and provenance of digital property; and 2) Storage and exchange of digital property with electronic contracts, automated transactions and autonomous agents

Digital Identity, including: 1) Technology and architecture for autonomy and control of self-sourced digital identity and personal data; and 2) Using individual identity for valid, verifiable login to apps or services and for providing legal acknowledgement, assent or authorization.

Digital Contracts, including 1) Integrating ordinary digital contracts and blockchain "smart contracts" in automated transactions by individuals or businesses; and 2) Standard open-web stack design patterns for executing multiple digital signatures and electronic notarization on digital legal contracts.

For more info and to apply, see: law.MIT.edu/Computational-Law-Course

Sponsor(s): Media Arts and Sciences
Contact: Dazza Greenwood, E15-449, 617.500.3644, DAZZA@MEDIA.MIT.EDU

---

Learning and Workshop Sessions

Jan/23	Mon	02:00PM-06:00PM	E15-341, MIT Media L, Primarily Learning/Discussing
Jan/24	Tue	02:00PM-06:00PM	E15-341, MIT Media L, Primarily Building/Exploring
Jan/26	Thu	06:30PM-11:30PM	TBD, Bonus ABA Hackathon Session
Jan/30	Mon	02:00PM-06:00PM	E15-341, MIT Media L, Primarily Learning/Discussing
Jan/31	Tue	02:00PM-06:00PM	E15-341, MIT Media L, Primarily Building/Exploring

For more information, please see: law.MIT.edu/Computational-Law-Course

Note: The Dates/Times and Place are Subject to Change

Professor Jonathan Askin - Professor of law, Dazza Greenwood, JD - Visiting Scientist, MIT Media Lab

Dr. Thomas R. Consi, Research Education Specialist, Sea Grant

Jan/09	Mon	01:00PM-04:00PM	5-007, bring notebook and pen to class
Jan/10	Tue	01:00PM-04:00PM	5-007, bring notebook and pen to class
Jan/11	Wed	01:00PM-04:00PM	5-007, bring notebook and pen to class
Jan/12	Thu	01:00PM-04:00PM	5-007, bring notebook and pen to class
Jan/13	Fri	01:00PM-04:00PM	5-007, bring notebook and pen to class

Enrollment: Limited: Advance sign-up required
Sign-up by 01/06
Limited to 16 participants
Attendance: Participants must attend all sessions
Prereq: Curiosity

Pond scum is a derogatory term derived from the notion that green slime on the surface of stagnant water is something disgusting - nothing could be further from the truth! Pond scum is in fact a Lilliputian world inhabited by an array of amazing creatures. This course is about these creatures: who they are, how they work, and the challenges they face living in micro-scale fluid environments. 

Topics:

The Light Microscope: Its design and operation. Micro-imaging and video techniques.

The Micro-Environment:  At tiny scales fluid flow is reversible, swimming creatures can stop instantly, and drag is not affected by shape. Learn the basic physics of fluid and flow at micro-scales and how microorganisms are adapted to live in this strange, counterintuitive world.

Diversity of Microorganisms:  Observe and learn to identify a diverse range of microorganisms. Some row with thousands of tiny hairs, some corkscrew through water, and some simply flow in whatever direction they please!

Measuring Microorganisms:  Measure the size and speed of microorganisms and estimate the forces they exert to crawl or swim. Perform experiments to see how they create flow fields for movement, sensing and feeding. Observe their behaviors and speculate on their “cognitive” abilities.

Bio-Inspired Micro-Robots:  At the end of the class, you will be challenged to design an aquatic micro-robot inspired by our exploration of live microorganisms.       

Sponsor(s): MIT-SUTD Collaboration
Contact: Dr. Thomas R. Consi, consi@mit.edu

Jose Gomez-Marquez, Little Devices Lab

Enrollment: Unlimited: Advance sign-up required
Sign-up by 01/10
Attendance: 1st and 2nd class mandatory (safety + drone intro) email instructor
Prereq: None

BUILD A DRONE // BUILD A DRONE RACECOURSE

Class is full.

Students will learn how to design, build and assemble their own FPV racing drone and work in teams to create a life size drone racing course. Advanced topics such as racing flight behavior, pop up architectural and infrastructure design, and sensor systems will be explored. The instructors will be drone racing professionals and MIT researchers.

In 2014, hobbyists in France decided to attach front-facing cameras and LEDs to their fast-flying drones, racing them “Star Wars-style” through the forest earning them more then 3M YouTube views. Drone racing has since expanded globally and is now featured in mainstream sports on ESPN. The nascent sport still offers plenty of opportunity for vehicle & course design innovation.

Students will learn how to build, fly, and race their own mini FPV racing drone, and work in teams to design, build and test a 1:1 scale drone racing course using 3-D printed components, architectural design, and sensor systems for pilot feedback.

Learn about the different obstacle courses that drone pilots before us have built, then improve upon and build your own. We will explore how different geometries, materials, and arrangements of obstacles affects the performance of a track. Each team’s racecourse will then be tested using biometrics as a measure of course capacity.

No prior flying experience is necessary. Basic safety training session is required. For more information, reach jfgm@mit.edu

Sponsor(s): MIT-SUTD International Design Center
Contact: Jose Gomez-Marquez, N52-373G, 617.674.7516, JFGM@MIT.EDU

---

Class 1

Jan/17	Tue	03:00PM-05:00PM	N52-373G
Jan/18	Wed	03:00PM-05:00PM	N52-373G
Jan/20	Fri	03:00PM-05:00PM	N52-373G
Jan/23	Mon	03:00PM-05:00PM	N52-373G
Jan/25	Wed	03:00PM-05:00PM	N25-373G
Jan/27	Fri	03:00PM-05:00PM	N52-373G

Sertac Karaman, Professor, Aero/Astro, Michael Boulet, Ken Gregson, Owen Guldner

Enrollment: Limited: Advance sign-up required
Sign-up by 01/02
Limited to 30 participants
Attendance: Participants must attend all sessions
Prereq: see description

Modern robots tend to operate at slow speeds in complex environments, limiting their utility in high-tempo applications. In this course you will push the boundaries of unmanned vehicle speed. Teams of 4-5 will develop dynamic autonomy software to race an RC car equipped with LIDAR, cameras, inertial sensors, and embedded processing around a large-scale, “real-world” course. Working from a baseline autonomy stack, teams will modify the software to increase platform velocity to the limits of stability. The course culminates with a timed competition to navigate the MIT tunnels. Classes will provide lectures on algorithms and lab time with instructor-assisted development. Must attend every class and plan on 6-10 hr/week of self-directed development.

Prereqs: Advanced undergraduates and graduates with some background in controls or robotics. Majors include AeroAstro, Mechanical, Ocean, and EECS. Students with a background in computer science with interest in robotics and controls may also effectively participate. Must have experience with software development. Past exposure to robotics algorithms and/or embedded programming will be useful. Email racecar-iap-course-subscribe@mit.edu with a brief description of your programming/robotics experience.

*This work is sponsored by the Dept. of the Air Force under Contract FA8721-05-C-0002.  Opinions, interpretations, conclusions and recommendations are those of the author and are not necessarily endorsed by the U.S. Government.

Sponsor(s): Electrical Engineering and Computer Science, Lincoln Laboratory, MIT-SUTD Collaboration
Contact: Owen Guldner, racecar-iap-course-subscribe@mit.edu

---

Jan/09	Mon	01:00PM-05:00PM	32-081
Jan/11	Wed	01:00PM-05:00PM	32-081
Jan/13	Fri	01:00PM-05:00PM	32-081
Jan/18	Wed	01:00PM-05:00PM	32-081
Jan/20	Fri	01:00PM-05:00PM	32-081

Students should be prepared to put in significant time outside of scheduled class hours (approx. 6-10 hours each week)

Fady Saad SM '13, Co-founder and Director of Partnerships, MassRobotics

Jan/17

Tue

06:00PM-07:00PM

32-124

Enrollment: Unlimited: Advance sign-up required

Alumnus Fady Saad SM '13 will speak about his journey from North Africa and Europe working in Fortune 500 companies to MIT, startups, robotics, and finally MassRobotics located here in Cambridge. Other topics include Fady's research, his theory on startups lifecycle, and the rise of new forms of innovation spaces.

Register today!

Sponsor(s): Alumni Association
Contact: Elena Byrne, W98-206C, 617 252-1143, EBYRNE@MIT.EDU

Henry Lieberman, Research Scientist, CSAIL, Christopher Fry

Jan/18

Wed

03:00PM-05:00PM

24-615

Enrollment: Unlimited: No advance sign-up
Prereq: none

Indeed, why can't we? Why do we have war? Poverty? What can we do
about it? Will technological progress result in robots destroying
humanity? Will automation take all our jobs? Will there be ecological
disaster?  What's the future of government, industry, education,
transportation, justice?

We'll show you a simple mathematical, psychological, and evolutionary
model that explains why people get sometimes sucked into doing bad
stuff, even if they're not bad people. We'll also explain how new
technology, especially AI and 3D printing, can enable a more just,
prosperous, and more cooperative society. Young people now have an
opportunity to rethink government, the economy, education, and all of
our institutions. Let's do it!

Feeling frustrated about your new President and the process that got
him there? Can technology help? Yes.

Sponsor(s): Experimental Study Group
Contact: Henry Lieberman, 32G-475, (617) 500-5267, lieber@MEDIA.MIT.EDU