This paper presents a tree-to-tree transduction method for sentence compression. Our model is based on synchronous tree substitution grammar, a formalism that allows local distortion of the tree topology and can thus naturally capture structural mismatches. We describe an algorithm for decoding in this framework and show how the model can be trained discriminatively within a large margin framework. Experimental results on sentence compression bring significant improvements over a state-of-the-art model.

This paper presents a new method for inferring the semantic properties of documents by leveraging free-text keyphrase annotations. Such annotations are becoming increasingly abundant due to the recent dramatic growth in semi-structured, user-generated online content. One especially relevant domain is product reviews, which are often annotated by their authors with pros/cons keyphrases such as ``a real bargain'' or ``good value.'' These annotations are representative of the underlying semantic properties; however, unlike expert annotations, they are noisy: lay authors may use different labels to denote the same property, and some labels may be missing. To learn using such noisy annotations, we find a hidden paraphrase structure which clusters the keyphrases. The paraphrase structure is linked with a latent topic model of the review texts, enabling the system to predict the properties of unannotated documents and to effectively aggregate the semantic properties of multiple reviews. Our approach is implemented as a hierarchical Bayesian model with joint inference. We find that joint inference increases the robustness of the keyphrase clustering and encourages the latent topics to correlate with semantically meaningful properties. Multiple evaluations demonstrate that our model substantially outperforms alternative approaches for summarizing single and multiple documents into a set of semantically salient keyphrases.

In this paper we explore avenues for improving the reliability of dimensionality reduction methods such as Non-Negative Matrix Factorization (NMF) as interpretive exploratory data analysis tools. We first explore the difficulties of the optimization problem underlying NMF, showing for the first time that non-trivial NMF solutions always exist and that the optimization problem is actually convex, by using the theory of Completely Positive Factorization. We subsequently explore four novel approaches to finding globally-optimal NMF solutions using various ideas from convex optimization. We then develop a new method, isometric NMF (isoNMF), which preserves non-negativity while also providing an isometric embedding, simultaneously achieving two properties which are helpful for interpretation. Though it results in a more difficult optimization problem, we show experimentally that the resulting method is scalable and even achieves more compact spectra than standard NMF.

Handling missing values in training datasets for constructing learning models or extracting useful information is considered to be an important research task in data mining and knowledge discovery in databases. In recent years, lot of techniques are proposed for imputing missing values by considering attribute relationships with missing value observation and other observations of training dataset. The main deficiency of such techniques is that, they depend upon single approach and do not combine multiple approaches, that why they are less accurate. To improve the accuracy of missing values imputation, in this paper we introduce a novel partial matching concept in association rules mining, which shows better results as compared to full matching concept that we described in our previous work. Our imputation technique combines the partial matching concept in association rules with k-nearest neighbor approach. Since this is a hybrid technique, therefore its accuracy is much better than as compared to those techniques which depend upon single approach. To check the efficiency of our technique, we also provide detail experimental results on number of benchmark datasets which show better results as compared to previous approaches.

The quality of training data for knowledge discovery in databases (KDD) and data mining depends upon many factors, but handling missing values is considered to be a crucial factor in overall data quality. Today real world datasets contains missing values due to human, operational error, hardware malfunctioning and many other factors. The quality of knowledge extracted, learning and decision problems depend directly upon the quality of training data. By considering the importance of handling missing values in KDD and data mining tasks, in this paper we propose a novel Hybrid Missing values Imputation Technique (HMiT) using association rules mining and hybrid combination of k-nearest neighbor approach. To check the effectiveness of our HMiT missing values imputation technique, we also perform detail experimental results on real world datasets. Our results suggest that the HMiT technique is not only better in term of accuracy but it also take less processing time as compared to current best missing values imputation technique based on k-nearest neighbor approach, which shows the effectiveness of our missing values imputation technique.

We study the distributions of the LASSO, SCAD, and thresholding estimators, in finite samples and in the large-sample limit. The asymptotic distributions are derived for both the case where the estimators are tuned to perform consistent model selection and for the case where the estimators are tuned to perform conservative model selection. Our findings complement those of Knight and Fu (2000) and Fan and Li (2001). We show that the distributions are typically highly nonnormal regardless of how the estimator is tuned, and that this property persists in large samples. The uniform convergence rate of these estimators is also obtained, and is shown to be slower than 1/root(n) in case the estimator is tuned to perform consistent model selection. An impossibility result regarding estimation of the estimators' distribution function is also provided.

We describe a preliminary investigation into learning a Chess player's style from game records. The method is based on attempting to learn features of a player's individual evaluation function using the method of temporal differences, with the aid of a conventional Chess engine architecture. Some encouraging results were obtained in learning the styles of two recent Chess world champions, and we report on our attempt to use the learnt styles to discriminate between the players from game records by trying to detect who was playing white and who was playing black. We also discuss some limitations of our approach and propose possible directions for future research. The method we have presented may also be applicable to other strategic games, and may even be generalisable to other domains where sequences of agents' actions are recorded.

Subspace clustering has gained increasing popularity in the analysis of gene expression data. Among subspace cluster models, the recently introduced order-preserving sub-matrix (OPSM) has demonstrated high promise. An OPSM, essentially a pattern-based subspace cluster, is a subset of rows and columns in a data matrix for which all the rows induce the same linear ordering of columns. Existing OPSM discovery methods do not scale well to increasingly large expression datasets. In particular, twig clusters having few genes and many experiments incur explosive computational costs and are completely pruned off by existing methods. However, it is of particular interest to determine small groups of genes that are tightly coregulated across many conditions. In this paper, we present KiWi, an OPSM subspace clustering algorithm that is scalable to massive datasets, capable of discovering twig clusters and identifying negative as well as positive correlations. We extensively validate KiWi using relevant biological datasets and show that KiWi correctly assigns redundant probes to the same cluster, groups experiments with common clinical annotations, differentiates real promoter sequences from negative control sequences, and shows good association with cis-regulatory motif predictions.

The class of chain event graph models is a generalisation of the class of discrete Bayesian networks, retaining most of the structural advantages of the Bayesian network for model interrogation, propagation and learning, while more naturally encoding asymmetric state spaces and the order in which events happen. In this paper we demonstrate how with complete sampling, conjugate closed form model selection based on product Dirichlet priors is possible, and prove that suitable homogeneity assumptions characterise the product Dirichlet prior on this class of models. We demonstrate our techniques using two educational examples.

This paper applies machine learning techniques to student modeling. It presents a method for discovering high-level student behaviors from a very large set of low-level traces corresponding to problem-solving actions in a learning environment. Basic actions are encoded into sets of domain-dependent attribute-value patterns called cases. Then a domain-independent hierarchical clustering identifies what we call general attitudes, yielding automatic diagnosis expressed in natural language, addressed in principle to teachers. The method can be applied to individual students or to entire groups, like a class. We exhibit examples of this system applied to thousands of students' actions in the domain of algebraic transformations.