Saturday, June 4, 2016

Claude Shannon Centenary and the 1-bit Maze (A-Maze-ing Dash Challenge)

The Claude Shannon Centenary 2016 in Hong Kong is a series of events held in Hong Kong to mark the life and legacy of Claude Shannon who was a visionary pioneer in the fields of computing, information theory, communications and machine intelligence. I chaired a seminar on quantum information theory given by the 2016 Shannon Award Lecturer Alexander Holevo on May 6th. Also, a Claude Shannon Centenary workshop was organised on May 19th by Professor Raymond Yeung at the CUHK Institute of Network Coding, in which I gave a talk titled "To prove or to disprove: information inequalities and sparse optimization". 

I also delivered this as an invited talk at Tsinghua University in Beijing on May 14th ("纪念Shannon诞辰一百周年学术论坛"). My talk was about using linear programming and cloud computing to automatically generate a mathematical proof or counterexample to Shannon-type inequalities in information theory. Fancy that! Shannon's Master thesis that engendered the birth of digital logic computing in 1938 and his seminal work on information theory in 1948, and these two seemingly-disparate subjects converge!

Next, a local outreach activity Computer Science Challenge was held on 21 May at the City University of Hong Kong for 180 middle school children (62 upper primary and 118 secondary school students) who formed teams to compete in three digital game challenges. We also had educational exhibits of Claude Shannon and replicas of some of his fun and thought-provoking robotic machines —Rubik-cube manipulators using Lego —on display. 

One of the CS Challenge's tasks was to program a robot to solve a maze and in the process to let the younger generation know more about Claude Shannon (recall that Shannon's Theseus mouse was the very first digitally-programmed maze-solver). The maze for secondary school students was designed with Shannon's entropy - a coin flip decides one of two possible maze entrances, i.e., the robot starts from the left entrance with a Head or otherwise from the right entrance with a Tail. The left and right entrances entail a left and right corner respectively, and the two passages meet at the intersection of a corridor to the exit. Now, a fair coin flip generates one bit of informationIn this way, as the coin is flipped only after the robot has been programmed, this 1-bit uncertainty prevents the participants from hard-coding the robot's movement (i.e., no dead reckoning); The robot has to hit an obstacle (i.e., the corner wall) and navigate its way by trial and error. Check out more pictures and the video below on the 1-bit maze. What a memorable and a-maze-ingly fun 2016 CS Challenge we had!

No comments:

Post a Comment