Jiaping Zhao

Learn More
Time series classification (TSC) arises in many fields and has a wide range of applications. Here, we adopt the bag-of-words (BoW) framework to classify time series. Our algorithm first samples local subsequences from time series at feature-point locations when available. It then builds local descriptors, and models their distribution by Gaussian mixture(More)
We presented a novel procedure to extract ground road networks from airborne LiDAR data. First point clouds were separated into ground and non-ground parts, and ground roads were to be extracted from ground planes. Then, buildings and trees were distinguished in an energy minimization framework after incorporation of two new features. The separation(More)
Continual Learning in artificial neural networks suffers from interference and forgetting when different tasks are learned sequentially. This paper introduces the Active Long Term Memory Networks (A-LTM), a model of sequential multitask deep learning that is able to maintain previously learned association between sensory input and behavioral output while(More)
This paper describes an unsupervised approach for efficient extraction of grid-structured urban roads from airborne LIDAR data. Technically, the approach consists of three major components: 1) terrain separation from DSM and classification of ground features, 2) road centerline extraction from generated road candidates images, and 3) completion and(More)
Predicting where humans will fixate in a scene has many practical applications. Biologically-inspired saliency models decompose visual stimuli into feature maps across multiple scales, and then integrate different feature channels, e.g., in a linear, MAX, or MAP. However, to date there is no universally accepted feature integration mechanism. Here, we(More)
Despite significant recent progress, the best available computer vision algorithms still lag far behind human capabilities, even for recognizing individual discrete objects under various poses, illuminations, and backgrounds. Here we present a new approach to using object pose information to improve deep network learning. While existing large-scale(More)
The complete mitochondrial genome (mtDNA) of Peking duck (Anas platyrhychos) was determined. The entire genome was 16,604 bp in length. Similar to the typical mtDNA of vertebrates, it contained 37 genes (13 protein-coding genes, 2 rRNA genes, 22 tRNA genes) and a non-coding region (D-loop). The characteristics of the mitochondrial genome was analyzed and(More)
We propose a novel univariate time series decomposition algorithm to partition temporal sequences into homogeneous segments. Unlike most existing temporal segmentation approaches, which generally build statistical models of temporal observations and then detect change points using inference or hypothesis testing techniques, our algorithm requires no domain(More)
Dynamic Time Warping (DTW) is an algorithm to align temporal sequences with possible local non-linear distortions, and has been widely applied to audio, video and graphics data alignments. DTW is essentially a point-to-point matching method under some boundary and temporal consistency constraints. Although DTW obtains a global optimal solution, it does not(More)