arXiv subjects

Yu-Chuan Su

Publications and source records attributed to Yu-Chuan Su.

21 records · Page 2Linked to original sources

arXiv2016

Leaving Some Stones Unturned: Dynamic Feature Prioritization for Activity Detection in Streaming Video

Current approaches for activity recognition often ignore constraints on computational resources: 1) they rely on extensive feature computation to obtain rich descriptors on all frames, and 2) they assume batch-mode access to the entire test video at once. We propose a new active approach to activity recognition that prioritizes "what to compute when" in order to make timely predictions. The main idea is to learn a policy that dynamically schedules the sequence of features to compute on selected frames of a given test video. In contrast to traditional static feature selection, our approach continually re-prioritizes computation based on the accumulated history of observations and accounts for the transience of those observations in ongoing video. We develop variants to handle both the batch and streaming settings. On two challenging datasets, our method provides significantly better accuracy than alternative techniques for a wide range of computational budgets.

cs.CV↗

arXiv2015

Transfer Learning for Video Recognition with Scarce Training Data for Deep Convolutional Neural Network

Unconstrained video recognition and Deep Convolution Network (DCN) are two active topics in computer vision recently. In this work, we apply DCNs as frame-based recognizers for video recognition. Our preliminary studies, however, show that video corpora with complete ground truth are usually not large and diverse enough to learn a robust model. The networks trained directly on the video data set suffer from significant overfitting and have poor recognition rate on the test set. The same lack-of-training-sample problem limits the usage of deep models on a wide range of computer vision problems where obtaining training data are difficult. To overcome the problem, we perform transfer learning from images to videos to utilize the knowledge in the weakly labeled image corpus for video recognition. The image corpus help to learn important visual patterns for natural images, while these patterns are ignored by models trained only on the video corpus. Therefore, the resultant networks have better generalizability and better recognition rate. We show that by means of transfer learning from image to video, we can learn a frame-based recognizer with only 4k videos. Because the image corpus is weakly labeled, the entire learning process requires only 4k annotated instances, which is far less than the million scale image data sets required by previous works. The same approach may be applied to other visual recognition tasks where only scarce training data is available, and it improves the applicability of DCNs in various computer vision problems. Our experiments also reveal the correlation between meta-parameters and the performance of DCNs, given the properties of the target problem and data. These results lead to a heuristic for meta-parameter selection for future researches, which does not rely on the time consuming meta-parameter search.

cs.CV↗

arXiv2013

Assessment of density functional methods with correct asymptotic behavior

Long-range corrected (LC) hybrid functionals and asymptotically corrected (AC) model potentials are two distinct density functional methods with correct asymptotic behavior. They are known to be accurate for properties that are sensitive to the asymptote of the exchange-correlation potential, such as the highest occupied molecular orbital energies and Rydberg excitation energies of molecules. To provide a comprehensive comparison, we investigate the performance of the two schemes and others on a very wide range of applications, including the asymptote problems, self-interaction-error problems, energy-gap problems, charge-transfer problems, and many others. The LC hybrid scheme is shown to consistently outperform the AC model potential scheme. In addition, to be consistent with the molecules collected in the IP131 database [Y.-S. Lin, C.-W. Tsai, G.-D. Li, and J.-D. Chai, J. Chem. Phys., 2012, 136, 154109], we expand the EA115 and FG115 databases to include, respectively, the vertical electron affinities and fundamental gaps of the additional 16 molecules, and develop a new database AE113 (113 atomization energies), consisting of accurate reference values for the atomization energies of the 113 molecules in IP131. These databases will be useful for assessing the accuracy of density functional methods.

physics.chem-ph↗