In this paper we formulate a novel AND/OR graph representation capable of describing the different configurations of deformable articulated objects such as horses. The representation makes use of the summarization principle so that lower level nodes in the graph only pass on summary statistics to the higher level nodes. The probability distributions are invariant to position, orientation, and scale. We develop a novel inference algorithm that combined a bottom-up process for proposing configurations for horses together with a top-down process for refining and validating these proposals. The strategy of surround suppression is applied to ensure that the inference time is polynomial in the size of input data.

The human vision system is able to effortlessly perceive both short-range and long-range motion patterns in complex dynamic scenes. Previous work has assumed that two different mechanisms are involved in processing these two types of motion. In this paper, we propose a hierarchical model as a unified framework for modeling both short-range and long-range motion perception. Our model consists of two key components: a data likelihood that proposes multiple motion hypotheses using nonlinear matching, and a hierarchical prior that imposes slowness and spatial smoothness constraints on the motion field at multiple scales. We tested our model on two types of stimuli, random dot kinematograms and multiple-aperture stimuli, both commonly used in human vision research. We demonstrate that the hierarchical model adequately accounts for human performance in psychophysical experiments.

We propose a learning framework, which is based on diffusionmethodology, that performs data fusion and anomalydetection in multi-dimensional time series data. Real lifeapplications and processes usually contain a large numberof sensors that generate parameters (features), where eachsensor collects partial information about the running process.These input sensors are fused to describe the behaviorof the whole process. The proposed data fusing algorithmis done in an hierarchial fashion: first it re-scales the inputsensors. Then, the re-formulated inputs are fused togetherby the application of the diffusion maps to reveal the nonlinearrelationships among them. This process constructsby embedding a low-dimensional description of the system.The embedding separates between sensors (parameters) thatcause stable and instable behavior of the system.This unsupervised algorithm first studies the system’sprofile from a training dataset by reducing its dimensions.Then, the coordinates of newly arrived data points are determinedby the application of multi-scale Gaussian approximation.To achieve this, an hierarchial processing of theincoming data is introduced.