Many data mining and data analysis techniques operate on dense matrices or complete tables of data. Real-world data sets, however, often contain unknown values. Even many classification algorithms that are designed to operate with missing values still exhibit deteriorated accuracy. One approach to handling missing values is to fill in (impute) the missing values. In this paper, we present a technique for unsupervised learning called Unsupervised Backpropagation (UBP), which trains a multi-layer perceptron to fit to the manifold sampled by a set of observed point-vectors. We evaluate UBP with the task of imputing missing values in datasets, and show that UBP is able to predict missing values with significantly lower sum-squared error than other collaborative filtering and imputation techniques. We also demonstrate with 24 datasets and 9 supervised learning algorithms that classification accuracy is usually higher when randomly-withheld values are imputed using UBP, rather than with other methods.

We propose a voted dual averaging method for online classification problems with explicit regularization. This method employs the update rule of the regularized dual averaging (RDA) method, but only on the subsequence of training examples where a classification error is made. We derive a bound on the number of mistakes made by this method on the training set, as well as its generalization error rate. We also introduce the concept of relative strength of regularization, and show how it affects the mistake bound and generalization performance. We experimented with the method using $\ell_1$ regularization on a large-scale natural language processing task, and obtained state-of-the-art classification performance with fairly sparse models.

The condensed nearest neighbor (CNN) algorithm is a heuristic for reducing the number of prototypical points stored by a nearest neighbor classifier, while keeping the classification rule given by the reduced prototypical set consistent with the full set. I present an upper bound on the number of prototypical points accumulated by CNN. The bound originates in a bound on the number of times the decision rule is updated during training in the multiclass perceptron algorithm, and thus is independent of training set size.

In this paper, a simple, general method of adding auxiliary stochastic neurons to a multi-layer perceptron is proposed. It is shown that the proposed method is a generalization of recently successful methods of dropout (Hinton et al., 2012), explicit noise injection (Vincent et al., 2010; Bishop, 1995) and semantic hashing (Salakhutdinov & Hinton, 2009). Under the proposed framework, an extension of dropout which allows using separate dropping probabilities for different hidden neurons, or layers, is found to be available. The use of different dropping probabilities for hidden layers separately is empirically investigated.

Discriminative structured prediction models have been widely used in many natural language processing tasks, but it is challenging to apply the methods to semantic parsing. In this paper, by introducing hybrid tree as a latent structure variable to close the gap between the input sentences and output representations, we formulate semantic parsing as a structured prediction problem, based on the latent variable perceptron model incorporated with a tree edit-distance loss as optimization criterion. The proposed approach maintains the advantage of a discriminative model in accommodating flexible combination of features and naturally incorporates an efficient decoding algorithm in learning and inference. Furthermore, in order to enhance the efficiency and accuracy of inference, we design an effective approach based on vector space model to extract a smaller subset of relevant MR productions for test examples. Experimental results on publicly available corpus show that our approach significantly outperforms previous systems.

Hierarchical classification (HC) plays an significant role in machine learning and data mining. However, most of the state-of-the-art HC algorithms suffer from high computational costs. To improve the performance of solving, we propose a stochastic perceptron (SP) algorithm in the large margin framework. In particular, a stochastic choice procedure is devised to decide the direction of next iteration. We prove that after finite iterations the SP algorithm yields a sub-optimal solution with high probability if the input instances are separable. For large-scale and high-dimensional data sets, we reform SP to the kernel version (KSP), which dramatically reduces the memory space needed. The KSP algorithm has the merit of low space complexity as well as low time complexity. The experimental results show that our KSP approach achieves almost the same accuracy as the contemporary algorithms on the real-world data sets, but with much less CPU running time.

It has been known for a long time that the classical spherical perceptrons can be used as storage memories. Seminal work of Gardner, \cite{Gar88}, started an analytical study of perceptrons storage abilities. Many of the Gardner's predictions obtained through statistical mechanics tools have been rigorously justified. Among the most important ones are of course the storage capacities. The first rigorous confirmations were obtained in \cite{SchTir02,SchTir03} for the storage capacity of the so-called positive spherical perceptron. These were later reestablished in \cite{TalBook} and a bit more recently in \cite{StojnicGardGen13}. In this paper we consider a variant of the spherical perceptron that operates as a storage memory but allows for a certain fraction of errors. In Gardner's original work the statistical mechanics predictions in this directions were presented sa well. Here, through a mathematically rigorous analysis, we confirm that the Gardner's predictions in this direction are in fact provable upper bounds on the true values of the storage capacity. Moreover, we then present a mechanism that can be used to lower these bounds. Numerical results that we present indicate that the Garnder's storage capacity predictions may, in a fairly wide range of parameters, be not that far away from the true values.

Perceptrons have been known for a long time as a promising tool within the neural networks theory. The analytical treatment for a special class of perceptrons started in seminal work of Gardner \cite{Gar88}. Techniques initially employed to characterize perceptrons relied on a statistical mechanics approach. Many of such predictions obtained in \cite{Gar88} (and in a follow-up \cite{GarDer88}) were later on established rigorously as mathematical facts (see, e.g. \cite{SchTir02,SchTir03,TalBook,StojnicGardGen13,StojnicGardSphNeg13,StojnicGardSphErr13}). These typically related to spherical perceptrons. A lot of work has been done related to various other types of perceptrons. Among the most challenging ones are what we will refer to as the discrete perceptrons. An introductory statistical mechanics treatment of such perceptrons was given in \cite{GutSte90}. Relying on results of \cite{Gar88}, \cite{GutSte90} characterized many of the features of several types of discrete perceptrons. We in this paper, consider a similar subclass of discrete perceptrons and provide a mathematically rigorous set of results related to their performance. As it will turn out, many of the statistical mechanics predictions obtained for discrete predictions will in fact appear as mathematically provable bounds. This will in a way emulate a similar type of behavior we observed in \cite{StojnicGardGen13,StojnicGardSphNeg13,StojnicGardSphErr13} when studying spherical perceptrons.

Recently, we proposed to transform the outputs of each hidden neuron in a multi-layer perceptron network to have zero output and zero slope on average, and use separate shortcut connections to model the linear dependencies instead. We continue the work by firstly introducing a third transformation to normalize the scale of the outputs of each hidden neuron, and secondly by analyzing the connections to second order optimization methods. We show that the transformations make a simple stochastic gradient behave closer to second-order optimization methods and thus speed up learning. This is shown both in theory and with experiments. The experiments on the third transformation show that while it further increases the speed of learning, it can also hurt performance by converging to a worse local optimum, where both the inputs and outputs of many hidden neurons are close to zero.

We propose a new type of hidden layer for a multilayer perceptron, and demonstrate that it obtains the best reported performance for an MLP on the MNIST dataset.