This paper proposes a new model, called condition-transforming variational autoencoder (CTVAE), to improve the performance of conversation response generation using conditional variational autoencoders (CVAEs). In conventional CVAEs , the prior distribution of latent variable z follows a multivariate Gaussian distribution with mean and variance modulated by the input conditions. Previous work found that this distribution tends to become condition-independent in practical application. In our proposed CTVAE model, the latent variable z is sampled by performing a non-linear transformation on the combination of the input conditions and the samples from a condition-independent prior distribution N(0,I). In our objective evaluations, the CTVAE model outperforms the CVAE model on fluency metrics and surpasses a sequence-to-sequence (Seq2Seq) model on diversity metrics. In subjective preference tests, our proposed CTVAE model performs significantly better than CVAE and Seq2Seq models on generating fluency, informative and topic relevant responses.

We present results for automated text categorization of the Reuters-810000 collection of news stories. Our experiments use the entire one-year collection of 810,000 stories and the entire subject index. We divide the data into monthly groups and provide an initial benchmark of text categorization performance on the complete collection. Experimental results show that efficient sparse-feature implementations of linear methods and decision trees, using a global unstemmed dictionary, can readily handle applications of this size. Predictive performance is approximately as strong as the best results for the much smaller older Reuters collections. Detailed results are provided over time periods. It is shown that a smaller time horizon does not diminish predictive quality, implying reduced demands for retraining when sample size is large.

We describe a lightweight learning method that induces an ensemble of decision-rule solutions for regression problems. Instead of direct prediction of a continuous output variable, the method discretizes the variable by k-means clustering and solves the resultant classification problem. Predictions on new examples are made by averaging the mean values of classes with votes that are close in number to the most likely class. We provide experimental evidence that this indirect approach can often yield strong results for many applications, generally outperforming direct approaches such as regression trees and rivaling bagged regression trees.

Decision tree induction is a prominent learning method, typically yielding quick results with competitive predictive performance. However, it is not unusual to find other automated learning methods that exceed the predictive performance of a decision tree on the same application. To achieve near-optimal classification results, resampling techniques can be employed to generate multiple decision-tree solutions. These decision trees are individually applied and their answers voted. The potential for exceptionally strong performance is counterbalanced by the substantial increase in computing time to induce many decision trees. We describe estimators of predictive performance for voted decision trees induced from bootstrap (bagged) or adaptive (boosted) resampling. The estimates are found by examining the performance of a single tree and its pruned subtrees over a single, training set and a large test set. Using publicly available collections of data, we show that these estimates are usually quite accurate, with occasional weaker estimates. The great advantage of these estimates is that they reveal the predictive potential of voted decision trees prior to applying expensive computational procedures.

We describe a machine learning method for predicting the value of a real-valued function, given the values of multiple input variables. The method induces solutions from samples in the form of ordered disjunctive normal form (DNF) decision rules. A central objective of the method and representation is the induction of compact, easily interpretable solutions. This rule-based decision model can be extended to search efficiently for similar cases prior to approximating function values. Experimental results on real-world data demonstrate that the new techniques are competitive with existing machine learning and statistical methods and can sometimes yield superior regression performance.