NIPS 2014 papers

(in nicer format than this)
maintained by @karpathy
Below every paper are TOP 100 most-occuring words in that paper and their color is based on LDA topic model with k = 7.
(It looks like 0 = graphical learning?, 1 = reinforcement learning, 2 = deep learning, 3 = non-parametrics?, 4 = matrix factorization?, 5 = neuroscience, 6 = optimization)
Toggle LDA topics to sort by: TOPIC0 TOPIC1 TOPIC2 TOPIC3 TOPIC4 TOPIC5 TOPIC6
Also note that you can filter by the day each poster appears: click on the day next to a paper to filter by that day.
Time--Data Tradeoffs by Aggressive Smoothing
John J. Bruer, Joel A. Tropp, Volkan Cevher, Stephen Becker


[number, maximal, random, apply, increase, fact, experiment, set, procedure] [algorithm, proposition, total, case, result, bound, optimal, order, phase, sequence] [work, based, learning, applied, geometry, achieve, accurate, additional] [sample, figure, parameter, average, measurement, supplemental, data, amount, gaussian] [recovery, matrix, sparse, vector, dimension, exact, condition, sparsity, norm, rank, suggests, siam, numerical, ambient, point, nonzero] [inverse, signal, panel, observed] [problem, smoothing, statistical, size, convex, cost, method, dual, rlip, normalized, linear, regularizer, solve, primal, computational, error, excess, smoothed, aggressive, constant, optimization, descent, regularized, assume, gradient, geometric, convergence, large, complexity, relative, min, arg]
Bounded Regret for Finite-Armed Structured Bandits
Tor Lattimore, Remi Munos


[number, asymptotically, structure, interval] [regret, arm, bound, algorithm, optimal, ucb, bandit, expected, return, ern, logarithmic, action, lower, case, proof, thompson, best, setting, lai, general, lemma, max, chosen, gap, cumulative, choose, optimistic, bubeck, suffers, depend, leung, learner, minimum, unknown, agrawal, tze, bastien, choosing, bounded, herbert, upper, met, studied, advertising, main] [structured, work, learning, bounding, unit] [log, figure, parameter, true, sampling, sampled, expectation, robbins] [improved, second, union, analysis, small] [depicted] [theorem, problem, assume, function, constant, large, example, machine, note, suppose, stochastic, complexity]
Best-Arm Identification in Linear Bandits
Marta Soare, Alessandro Lazaric, Remi Munos


[set, number, needed, identify, selection, consider, discrete, high] [arm, allocation, strategy, stopping, design, optimal, static, best, phase, bandit, oracle, algorithm, logn, conference, budget, bound, max, notice, case, reward, mab, adaptive, unknown, previous, pulling, suboptimal, hlb, study, space, discarding, criterion, expected, sequence, setting, gap, focus, difference, learner, studied, minimum] [learning, performance, well, additional] [sample, estimate, parameter, uncertainty, sampling, international, allows, figure] [condition, corresponding, cone, matrix, point] [directly, experimental, observed] [complexity, linear, problem, min, arg, empirical, direction, theory, implemented, objective, estimation, discard, term, quantity, corresponds, machine]
Learning to Optimize via Information-Directed Sampling
Dan Russo, Benjamin Van Roy


[set, random, ratio, entropy, consider, binary, paper, experiment, solution, high] [action, thompson, regret, optimal, ucb, algorithm, expected, bandit, reward, bound, probability, gain, upper, type, period, studied, customer, policy, chosen, online, assortment, cumulative, exploration, general, proposition, chooses, outcome, lower, van, russo, focus, dramatically, version, satisfy, search, minimizes, utility] [performance, work, learning, feature, applied, learn] [sampling, distribution, posterior, gaussian, bayesian, true, prior, figure, parameter] [product, knowledge, numerical, analysis, vector, establish, denote, greedy] [time, independent, mutual, drawn, presented] [linear, problem, optimization, bayes, simple, measure, theoretical, stochastic, strong, gradient, arg, computational, positive]
Efficient learning by implicit exploration in bandit problems with side observations
Tomáš Kocák, Gergely Neu, Michal Valko, Remi Munos


[graph, set, number, directed, consider, random, partial] [algorithm, learner, exploration, regret, action, observability, implicit, combinatorial, lemma, news, observe, setting, feedback, online, bandit, bound, side, alon, decision, auer, fpl, chosen, upper, access, mannor, shamir, neu, proof, adaptive, result, optimistic, total, guarantee, choose, main, clickthroughs, observes, vtt, previous] [learning, performance, based, propose, computer, proposed] [log, observation, scheme, distribution, figure, model, full, parameter, independence, estimate, prior, computationally, sampling, bias] [vector, requires, analysis, refer, running] [system, time, presented] [loss, problem, assume, theorem, optimization, framework, prove, variant]
Estimation with Norm Regularization
Arindam Banerjee, Sheng Chen, Farideh Fazayeli, Vidyashankar Sivakumar