We present an approach to semi-supervised learning based on an exponential family characterization. Our approach generalizes previous work on coupled priors for hybrid generative/discriminative models. Our model is more flexible and natural than previous approaches. Experimental results on several data sets show that our approach also performs better in practice. 

In machine learning problems, labeled data are often in short supply. One of the feasible solution for this problem is transfer learning. It can make use of the labeled data from other domain to discriminate those unlabeled data in the target domain. In this paper, we propose a transfer learning framework based on similarity matrix approximation to tackle such problems. Two practical algorithms are proposed, which are the label propagation and the similarity propagation. In these methods, we build a hybrid graph based on all available data. Then the information is transferred cross domains through alternatively constructing the similarity matrix for different part of the graph. Among all related methods, similarity propagation approach can make maximum use of all available similarity information across domains. This leads to more efficient transfer and better learning result. The experiment on real world text mining applications demonstrates the promise and effectiveness of our algorithms.

We propose a novel approach to context sensitive semantic smoothing by making use of an intermediate, "semantically light" representation for sentences, called Semantically Relatable Sequences (SRS). SRSs of a sentence are tuples of words appearing in the semantic graph of the sentence as linked nodes depicting dependency relations. In contrast to patterns based on consecutive words, SRSs make use of groupings of non-consecutive but semantically related words. Our experiments on TREC AP89 collection show that the mixture model of SRS translation model and Two Stage Language Model (TSLM) of Lafferty and Zhai achieves MAP scores better than the mixture model of MultiWord Expression (MWE) translation model and TSLM. Furthermore, a system, which for each test query selects either the SRS or the MWE mixture model based on better query MAP score, shows significant improvements over the individual mixture models.

Despite the success of Gaussian processes (GPs) in modelling spatial stochastic processes, dealing with large datasets is still challenging. The problem arises by the need to invert a potentially large covariance matrix during inference. In this paper we address the complexity problem by constructing a new stationary covariance function (Mercer kernel) that naturally provides a sparse covariance matrix. The sparseness of the matrix is defined by hyper-parameters optimised during learning. The new covariance function enables exact GP inference and performs comparatively to the squared-exponential one, at a lower computational cost. This allows the application of GPs to large-scale problems such as ore grade prediction in mining or 3D surface modelling. Experiments show that using the proposed covariance function, very sparse covariance matrices are normally obtained which can be effectively used for faster inference and less memory usage.

Branch and bound is an effective technique for solving constraint optimization problems (COP’s). However, its search space expands very rapidly as the domain sizes of the problem variables grow. In this paper, we present an algorithm that clusters the values of a variable’s domain into sets. Branch and bound can then branch on these sets of values rather than on individual values, thereby reducing the branching factor of its search space. The aim of our clustering algorithm is to construct a collection of sets such that branching on these sets will still allow effective bounding. In conjunction with the reduced branching factor, the size of the explored search space is thus significantly reduced. We test our method and show empirically that it can yield significant performance gains over existing stateof- the-art techniques.

Canonical correlation analysis (CCA) and partial least squares (PLS) are well-known techniques for feature extraction from two sets of multi-dimensional variables. The fundamental difference between CCA and PLS is that CCA maximizes the correlation while PLS maximizes the covariance. Although both CCA and PLS have been applied successfully in various applications, the intrinsic relationship between them remains unclear. In this paper, we attempt to address this issue by showing the equivalence relationship between CCA and orthonormalized partial least squares (OPLS), a variant of PLS. We further extend the equivalence relationship to the case when regularization is employed for both sets of variables. In addition, we show that the CCA projection for one set of variables is independent of the regularization on the other set of variables. We have performed experimental studies using both synthetic and real data sets and our results confirm the established equivalence relationship. The presented analysis provides novel insights into the connection between these two existing algorithms as well as the effect of the regularization.

Detecting abnormal activities from sensor readings is an important research problem in activity recognition. A number of different algorithms have been proposed in the past to tackle this problem. Many of the previous state-based approaches suffer from the problem of failing to decide the appropriate number of states, which are difficult to find through a trial and-error approach, in real-world applications. In this paper, we propose an accurate and flexible framework for abnormal activity recognition from sensor readings that involves less human tuning of model parameters. Our approach first applies a Hierarchical Dirichlet Process Hidden Markov Model (HDP-HMM), which supports an infinite number of states, to automatically find an appropriate number of states. We incorporate a Fisher Kernel into the One-Class Support Vector Machine (OCSVM) model to filter out the activities that are likely to be normal. Finally, we derive an abnormal activity model from the normal activity models to reduce false positive rate in an unsupervised manner. Our main contribution is that our proposed HDP-HMM models can decide the appropriate number of states automatically, and that by incorporating a Fisher Kernel into the OCSVM model, we can combine the advantages from generative model and discriminative model. We demonstrate the effectiveness of our approach by using several real-world datasets to test our algorithm’s performance.

Multi-class problems have a richer structure than binary classification problems. Thus, they can potentially improve their performance by exploiting the relationship among class labels. While for the purposes of providing an automated classification result this class structure does not need to be explicitly unveiled, for human level analysis or interpretation this is valuable. We develop a multi-class large margin classifier that extracts and takes advantage of class relationships. We provide a bi-convex formulation that explicitly learns a matrix that captures these class relationships and is de-coupled from the feature weights. Our representation can take advantage of the class structure to compress the model by reducing the number of classifiers employed, maintaining high accuracy even with large compression. In addition, we present an efficient formulation in terms of speed and memory.

Recently graph based dimensionality reduction has received a lot of interests in many fields of information processing. Central to it is a graph structure which models the geometrical and discriminant structure of the data manifold. When label information is available, it is usually incorporated into the graph structure by modifying the weights between data points. In this paper, we propose a novel dimensionality reduction algorithm, called Constrained Graph Embedding, which considers the label information as additional constraints. Specifically, we constrain the space of the solutions that we explore only to contain embedding results that are consistent with the labels. Experimental results on two real life data sets illustrate the effectiveness of our proposed method.

Open-ended language communication remains an enormous challenge for autonomous robots. This paper argues that the notion of a language strategy is the appropriate vehicle for addressing this challenge. A language strategy packages all the procedures that are necessary for playing a language game. We present a specific example of a language strategy for playing an Action Game  in which one robot asks another robot to take on a body posture (such as stand or sit), and show how it effectively allows a population of agents to self-organise a perceptually grounded ontology and a lexicon from scratch, without any human intervention. Next, we show how a new language strategy can arise by exaptation from an existing one, concretely, how the body posture strategy can be exapted to a strategy for playing language games about the spatial position of objects (as in "the bottle stands on the table").