We develop an iterative subsampling approach to improve the computational efficiency of our previous work on solution path clustering (SPC). The SPC method achieves clustering by concave regularization on the pairwise distances between cluster centers. This clustering method has the important capability to recognize noise and to provide a short path of clustering solutions; however, it is not sufficiently fast for big datasets. Thus, we propose a method that iterates between clustering a small subsample of the full data and sequentially assigning the other data points to attain orders of magnitude of computational savings. The new method preserves the ability to isolate noise, includes a solution selection mechanism that ultimately provides one clustering solution with an estimated number of clusters, and is shown to be able to extract small tight clusters from noisy data. The method's relatively minor losses in accuracy are demonstrated through simulation studies, and its ability to handle large datasets is illustrated through applications to gene expression datasets. An R package, SPClustering, for the SPC method with iterative subsampling is available at http://www.stat.ucla.edu/~zhou/Software.html.

This work analyzes centered binary Restricted Boltzmann Machines (RBMs) and binary Deep Boltzmann Machines (DBMs), where centering is done by subtracting offset values from visible and hidden variables. We show analytically that (i) centering results in a different but equivalent parameterization for artificial neural networks in general, (ii) the expected performance of centered binary RBMs/DBMs is invariant under simultaneous flip of data and offsets, for any offset value in the range of zero to one, (iii) centering can be reformulated as a different update rule for normal binary RBMs/DBMs, and (iv) using the enhanced gradient is equivalent to setting the offset values to the average over model and data mean. Furthermore, numerical simulations suggest that (i) optimal generative performance is achieved by subtracting mean values from visible as well as hidden variables, (ii) centered RBMs/DBMs reach significantly higher log-likelihood values than normal binary RBMs/DBMs, (iii) centering variants whose offsets depend on the model mean, like the enhanced gradient, suffer from severe divergence problems, (iv) learning is stabilized if an exponentially moving average over the batch means is used for the offset values instead of the current batch mean, which also prevents the enhanced gradient from diverging, (v) centered RBMs/DBMs reach higher LL values than normal RBMs/DBMs while having a smaller norm of the weight matrix, (vi) centering leads to an update direction that is closer to the natural gradient and that the natural gradient is extremly efficient for training RBMs, (vii) centering dispense the need for greedy layer-wise pre-training of DBMs, (viii) furthermore we show that pre-training often even worsen the results independently whether centering is used or not, and (ix) centering is also beneficial for auto encoders.

Conflict-Based Search (CBS) and its enhancements, Meta-Agent CBS and bypassing conflicts are amongst the strongest newly introduced algorithms for Multi-Agent Path Finding. This paper introduces two new improvements to CBS and incorporates them into a coherent, improved version of CBS, namely ICBS. Experimental results show that each of these improvements further reduces the runtime over the existing CBS-based approaches. When all improvements are combined, an even larger improvement is achieved, producing state-of-the art results for a number of domains.

The investigation and development of new methods from diverse perspectives to shed light on portfolio choice problems has never stagnated in financial research. Recently, multi-armed bandits have drawn intensive attention in various machine learning applications in online settings. The tradeoff between exploration and exploitation to maximize rewards in bandit algorithms naturally establishes a connection to portfolio choice problems. In this paper, we present a bandit algorithm for conducting online portfolio choices by effectually exploiting correlations among multiple arms. Through constructing orthogonal portfolios from multiple assets and integrating with the upper confidence bound bandit framework, we derive the optimal portfolio strategy that represents the combination of passive and active investments according to a risk-adjusted reward function. Compared with oft-quoted trading strategies in finance and machine learning fields across representative real-world market datasets, the proposed algorithm demonstrates superiority in both risk-adjusted return and cumulative wealth.

Expectation maximization (EM) has recently been shown to be an efficient algorithm for learning finite-state controllers (FSCs) in large decentralized POMDPs (Dec-POMDPs). However, current methods use fixed-size FSCs and often converge to maxima that are far from the optimal value. This paper considers a variable-size FSC to represent the local policy of each agent. These variable-size FSCs are constructed using a stick-breaking prior, leading to a new framework called decentralized stick-breaking policy representation (Dec-SBPR). This approach learns the controller parameters with a variational Bayesian algorithm without having to assume that the Dec-POMDP model is available. The performance of Dec-SBPR is demonstrated on several benchmark problems, showing that the algorithm scales to large problems while outperforming other state-of-the-art methods.

Large multilayer neural networks trained with backpropagation have recently achieved state-of-the-art results in a wide range of problems. However, using backprop for neural net learning still has some disadvantages, e.g., having to tune a large number of hyperparameters to the data, lack of calibrated probabilistic predictions, and a tendency to overfit the training data. In principle, the Bayesian approach to learning neural networks does not have these problems. However, existing Bayesian techniques lack scalability to large dataset and network sizes. In this work we present a novel scalable method for learning Bayesian neural networks, called probabilistic backpropagation (PBP). Similar to classical backpropagation, PBP works by computing a forward propagation of probabilities through the network and then doing a backward computation of gradients. A series of experiments on ten real-world datasets show that PBP is significantly faster than other techniques, while offering competitive predictive abilities. Our experiments also show that PBP provides accurate estimates of the posterior variance on the network weights.

We consider factoring low-rank tensors in the presence of outlying slabs. This problem is important in practice, because data collected in many real-world applications, such as speech, fluorescence, and some social network data, fit this paradigm. Prior work tackles this problem by iteratively selecting a fixed number of slabs and fitting, a procedure which may not converge. We formulate this problem from a group-sparsity promoting point of view, and propose an alternating optimization framework to handle the corresponding $\ell_p$ ($0

The time complexity of problems and algorithms, i.e., the scaling of the time required for solving a problem instance as a function of instance size, is of key interest in theoretical computer science and practical applications. In this context, propositional satisfiability (SAT) is one of the most intensely studied problems, and it is generally believed that solving SAT requires exponential time in the worst case. For more than two decades, random 3-SAT at the solubility phase transition has played a pivotal role in the theoretical and empirical investigation of SAT solving, and to this day, it is arguably the most prominent model for difficult SAT instances. Here, we study the empirical scaling of the running time of several prominent, high-performance SAT solvers on random 3-SAT instances from the phase transition region. After introducing a refined model for the location of the phase transition point, we show that the median running time of three incomplete, SLS-based solvers – WalkSAT/SKC, BalancedZ and probSAT – scales polynomially with instance size. An analogous analysis of three complete, DPLL-based solvers – kcnfs, march_hi and march_br – clearly indicates exponential scaling of median running time. Moreover, exponential scaling is witnessed for these DPLL-based solvers when solving only satisfiable and only unsatisfiable instances, and the respective scaling models for each solver differ mostly by a constant factor.

Cost-optimal planning (COP) uses action costs and asks for a minimum-cost plan. It is sometimes assumed that there is no harm in using actions with zero cost or rational cost. Classical complexity analysis does not contradict this assumption; planning is PSPACE-complete regardless of whether action costs are positive or non-negative, integer or rational. We thus apply parameterised complexity analysis to shed more light on this issue. Our main results are the following. COP is [W2]-complete for positive integer costs, i.e. it is no harder than finding a minimum-length plan, but it is paraNP-hard if the costs are non-negative integers or positive rationals. This is a very strong indication that the latter cases are substantially harder. Net-benefit planning (NBP) additionally assigns goal utilities and asks for a plan with maximum difference between its utility and its cost. NBP is paraNP-hard even when action costs and utilities are positive integers, suggesting that it is harder than COP. In addition, we also analyse a large number of subclasses, using both the PUBS restrictions and restricting the number of preconditions and effects.

Distributed word representations have a rising interest in NLP community. Most of existing models assume only one vector for each individual word, which ignores polysemy and thus degrades their effectiveness for downstream tasks. To address this problem, some recent work adopts multi-prototype models to learn multiple embeddings per word type. In this paper, we distinguish the different senses of each word by their latent topics. We present a general architecture to learn the word and topic embeddings efficiently, which is an extension to the Skip-Gram model and can model the interaction between words and topics simultaneously. The experiments on the word similarity and text classification tasks show our model outperforms state-of-the-art methods.