Head movement during scanning introduces artificial signal changes and impedes activation detection in fMRI studies. The head motion in fMRI acquired using slice-based Echo Planar Imaging (EPI) sequence can be estimated and compensated by aligning the images onto a reference volume through image registration. Registering EPI images volume by volume fails to consider head motion between slices, leading to biased head motion estimates. Slice-to-volume registration is used to estimate motion parameters for each slice by more accurately representing the image acquisition sequence. However, it is prone to image noise and geometric distortion, resulting in high variance estimates. In this work, we propose a Gaussian particle filter based head motion tracking algorithm to reduce the image misregistration errors. The algorithm models head motion by using a dynamic state space model (SSM) to model continuous slice acquisition thereby providing more accurate motion estimates and voxel position estimates. We demonstrate significant performance improvement of the proposed approach as compared to previous registration-only methods of head motion estimation.

The Recurrent temporal restricted Boltzmann machine (RTRBM) is a probabilistic model for temporal data, that has been shown to effectively capture both short and long-term dependencies in time-series. The topology of the RTRBM graphical model, however, assumes full connectivity between all the pairs of visible and hidden units, therefore ignoring the dependency structure between the different observations. Learning this structure has the potential to not only improve the prediction performance, but it can also reveal important patterns in the data. For example, given an econometric dataset, we could identify interesting dependencies between different market sectors; given a meteorological dataset, we could identify regional weather patterns. In this work we propose a new class of RTRBM, which explicitly uses a dependency graph to model the structure in the problem and to define the energy function. We refer to the new model as the structured RTRBM (SRTRBM). Our technique is related to methods such as graphical lasso, which are used to learn the topology of Gaussian graphical models. We also develop a spike-and-slab version of the RTRBM, and combine it with our method to learn structure in datasets with real valued observations. Our experimental results using synthetic and real datasets, demonstrate that the SRTRBM can improve the prediction performance of the RTRBM, particularly when the number of visible units is large and the size of the training set is small. It also reveals the structure underlying our benchmark datasets.

The use of semantic attributes in computer vision problems has been gaining increased popularity in recent years. Attributes provide an intermediate feature representation in between low-level features and the class categories, and offer several attractive properties, among which are improved learning of novel categories based on few examples, as well as allowing for zero-shot learning. However, the major caveat is that learning semantic attributes is a laborious task, requiring a significant amount of time and human intervention to provide labels. In order to address this issue, we propose a weakly supervised approach to learn mid-level features, where the only supervision is provided by the category classes of the training examples. We develop a novel extension of the restricted Boltzmann machine (RBM) with Beta-Bernoulli process priors. Unlike the standard RBM, our model uses the class labels to promote more efficient sharing of information by different categories. This tends to improve the generalization performance. By using semantic attributes for which annotations are available, we show that we can find correspondences between the mid-level features that we learn and the labeled attributes. Therefore, the mid-level features have distinct semantic characterization which is very similar to that given by the semantic attributes, even though their labeling was not used during the training process. Our experimental results in object recognition tasks show significant performance gains, outperforming methods which rely on manually labeled semantic attributes.

This paper is concerned with joint Bayesian endmember extraction and linear unmixing of hyperspectral images using a spatial prior on the abundance vectors. We hypothesize that hyperspectral images are composed of two types of regions. For the first type, the material proportions of adjacent pixels are similar and can be jointly characterized by a single vector, and in the second, neighboring pixels have very different abundances and are characterized by unique mixing proportions. Using this hypothesis we propose a new unmixing algorithm which simultaneously segments the image into such regions and performs unmixing. The experimental results show that the new algorithm can lead to improved MSE of both the extracted endmembers and the estimated abundances in low SNR cases.

A new class of features for wavelet-based texture classification is introduced using a new feature-weighting scheme adapted to non-Euclidean similarity measures. The feature extraction is based on the histogram of the local second moment estimates of the wavelet transform. It is shown that the bins' centers of such histograms should be scaled logarithmically rather than linearly. The distance between two texture features is measured using the x/sup 2/ similarity measure, weighted according to the feature's degree of dispersion within the training dataset. Classification experiments of the proposed approach using an orthonormal wavelet transform show improved classification results compared to presently available methods.