Jeopardy and Probability

posted in: Blog, CV STEM 2015 | 0

Michael Leventhal, Chief Nerdy Development Officer

July 21, 2015

We start each week of the Colonie de Vacances with something we call “Teach Back”; a review of the previous week’s lessons. The students actively demonstrate their understanding of the material, reinforcing what they have learned. This week took the form of a game of Jeopardy where teams of students selected questions at augmenting values in iNERDE’s core categories of Science, Technology, Engineering, and Mathematics and a fifth having to do life goals, social responsibility, and the overall role of STEM in society. The game format was great, fun and motivational … once the students understood it. None of them had ever seen the game of Jeopardy before and it took them some time to get it. There is a surprising amount of cultural information embedded in the structure of the game. The player strategy is also complex, requiring some sense of how to assess the probability of having the right answer against the value of question against one’s position in relation to the other teams. The great thing about the game format is that the students absorb its complex principles bit by bit, simply by seeking to improve the outcome. Each question is like a mini-experiment; one sees the result, whether positive or negative, and adjusts their strategy for the next round, eventually absorbing the principles that lead to an optimum result.

11721421_10155830172300416_1340878931_n

11774815_10155830172275416_1448224041_n

Our game of Jeopardy led to an impromptu lesson the next day on basic probability. We began simply with flipping a coin to convey the idea of equal probability and the random nature in a single instance of the coin falling on heads or tails. The kids gathering in a circle around me, excitedly yelling their guesses for each coin flip; we must have looked like an impromptu back-alley craps game. We then recorded the result for ten coin flips and verified that while one could have the same side of the coin appear on multiple experiments, after enough experiments the result arrives, as it happened, at exactly 50-50. I added a second coin and we discussed the probability of each combination of coins. Pas evident! (Not obvious) I showed that in enumerating the possible outcomes one could easily see the probability of each: heads-heads, heads-tails, tails-heads, and tails-tails each being one possibility of a total of four, if one didn’t consider heads-tails or tails-heads there are 2 possibilities out of 4 and so on. The main objective was to enable the students to grasp the idea of approximating the probability and being able to select the most probable outcome. To reinforce that we did one final exercise. We took several die and colored different number of sides one color or another. The objective was for the students to make a rough calculation if the probability was more or less for each combination of colors.

Finally, I wanted to make a combination between randomness, probability and our studies in computer science. I gave the example of a video game where the programmer must make decide how to make a character move and act. I explained that the software does the equivalent of a coin flip, or in the case where one wants one thing sometimes but something else more often, it is like using the die with an uneven distribution of colors.

I think these lessons gave the students the beginning of thinking in a mathematically precise way, and how mathematical thinking can be applied in every day life. The lessons were very exciting because they were delivered in the forms of games, with the opportunity for each student to form and express a theory about the possible outcomes and to see if their theory worked. And if we do not succeed in forming excellent future computer scientists (the probability is extremely high), our iNERDE students may become successful gamblers (the probability of that is very, very low).

Machine Learning on the way to Dakar

posted in: Blog | 0

Michael Leventhal, Chief Nerdy Development Officer

July 11, 2015

PARIS, FRANCE – I’m on my way to Dakar for the final week of iNERDE’s first STEM summer camp in Sénégal, held in cooperation with the Senegalese-American Bilingual School, SABS. I’ll be teaching the computer systems and computer science curriculum developed by iNERDE and helping to make our expansion to Senegal a success.

I’m travelling to Dakar by way of the northern franco-flemish city of Lille, France. I spent the last few days in Lille at the International Conference on Machine Learning, ICML, a gathering of the world’s top researchers in the science and mathematics behind artificial intelligence. Machine Learning is a revolution in the making, enabling computational devices to understand what they see, hear, and sense and to make decisions that can improve on human performance levels on the same tasks. One of the most dramatic examples of the use of this technology is self-driving cars; even I, as a computer scientist active in the field, did not realize how far along this technology is until a friend who works at Tesla took me for a ride down one of the busiest highways in Silicon Valley – without a driver.

I work for a company, Xilinx, that makes a kind of chip, an FPGA, that is used for machine learning. An FPGA is different from a CPU in being a flexible, reconfigurable, connected mesh of parallel computing elements – a bit like our brains, a flexible, reconfigurable, connected mesh of neurons. Parallel computing is the frontier of computer architecture, enabling us to create machines that are not only faster and more efficient than conventional CPUs at the stuff that computers do today, but also to create new uses of computers that can, like humans beings, deal with fuzzy information, make pretty good decisions based on what can be known, and can learn over time to make better decisions in the future. The old way of designing CPUs, the Von Neumann architecture, is about 75 years old. As exciting as the things we can do with machine learning are, for me, as a computer scientist, the most exciting thing is the fact that we are making an evolutionary leap in our understanding of how to build computers. Sometimes people worry that computers are getting too smart but it is actually human beings that are getting smarter, and have taught themselves to build exponentially on what we have learned. I felt in awe listening to artificial intelligence researchers at ICML sharing their profound knowledge and witnessing their passion for discovery.

I feel that my work for iNERDE is very related to my work on parallel computing. Both are ultimately about reaching the next level of human creativity. iNERDE is a passion for discovery of new educational paradigms. Like the Von Neumann architecture in computing, our modern educational systems have accomplished extraordinary things – but there is another level for us to aspire to. iNERDE, by choosing to begin its work in Africa, is also aiming at the next level in social evolution. We don’t accept the status quo of rich and poor nations. We think that every human being must have the right to the opportunity to develop her or his capacities, to create, and to contribute.

In the check-in queue at airport in Paris I felt myself already in Africa. It was chaos, of course, with everyone jostling to get into the line. It wasn’t aggressive, people were easy and friendly. I heard many languages all around, African languages I can’t recognize, parents speaking to their children in African-accented French and their kids responding in perfect Parisian French, and even Africans speaking unaccented American English. Many men and women were beautifully dressed in traditional African clothes.

At the security screening the agent looked at my ticket and said, “You’re going to Dakar! You’re going to love it, but it is really hot there! Where are you from? Oh, America! Los Angeles? I love America, I want to go to New York City and California and everywhere. Hey, cool mec! Welcome to Dakar!”