3. Finding the distribution that has maximal entropy over the set of linear constraints is equivalent to finding the I-projection of the uniform distribution onto the constraints. Who pioneered the geometric approach of I-projection? a) S.-i. Amari b) I. Csiszar c) N. Chentsov 4. The Kullback-Leibler divergence can be lower bounded using the variational metric. What is the generic name for designing such an inequality Cramer-Rao inequality Bhattacharyya inequality Pinsker inequality Bayes inequality 5. Both Renyi entropies (extensive) and Tsallis entropies (non-extensive) generalizing Shannon entropy, can be interpreted using a biparameter family of entropies. Mittal, and Sharma and Taneja independently published papers in which year? 1971 1975 1982 1988 6. Give a famous example of exponential families in statistical physics: Einstein model Botzlmann model Ising model Gibbs model
7. In physics, EM means electro-magnetism. But what does this acronym mean in statistics/computer science: empirical minimization emipirical maximization extended minimization expectation maximization
8. In statistical physics, what does a Gibbs distribution mean? a Markov Random Field a Conditional Random Field a Hidden Markov Model an exponential family
9. In 1957, E.T. Jaynes came up with two papers introducing the novel MaxEnt statistical inference scheme for thermodynamics. Jaynes started his PhD with Oppenheimer but aborted, he then restarted a PhD with Eugene Wigner. What are the first names of E. T.? Edgard Thompson Edwin Thompson Edgard Thomason Edwin Thomason
10. Quasi-arithmetic means were independently studied by Kolmogorov and Nagumo. What is the firstname of Nagumo in the original paper M. Nagumo, ¨Uber eine klasse der mittelwerte, Japan. J. Math., 7 (1930) 71–79. Mitio Mikiko Masaki