Mizzou IEEE CIS

Hi MU CISers,

Tomorrow, Alvey Brendan will present a seminar on "Deep Supervised and Contractive Neural Network for SAR Image Classification" at 2:00 PM (in order to avoid potential faculty meetings) in Ketcham Auditorium W1005.

Abstract of the paper:
The classification of a synthetic aperture radar (SAR) image is a significant yet challenging task, due to the presence of speckle noises and the absence of effective feature representation. Inspired by deep learning technology, a novel deep supervised and contractive neural network (DSCNN) for SAR image classification is proposed to overcome these problems. In order to extract spatial features, a multiscale patch-based feature extraction model that consists of gray level-gradient co- occurrence matrix, Gabor, and histogram of oriented gradient descriptors is developed to obtain primitive features from the SAR image. Then, to get discriminative representation of initial features, the DSCNN network that comprises four layers of supervised and contractive autoencoders is proposed to optimize features for classification. The supervised penalty of the DSCNN can capture the relevant information between features and labels, and the contractive restriction aims to enhance the locally invari- ant and robustness of the encoding representation. Consequently, the DSCNN is able to produce effective representation of sample features and provide superb predictions of the class labels. Moreover, to restrain the influence of speckle noises, a graph- cut-based spatial regularization is adopted after classification to suppress misclassified pixels and smooth the results. Experiments on three SAR data sets demonstrate that the proposed method is able to yield superior classification performance compared with some related approaches

Hi MU CISers,

Today, we will have paper discussion start with Chapter 4 of "Learning Deep Architectures for AI" at 3 pm in Ketcham Auditorium W1005.

Hi MU CISers,

Today, we will have paper discussion start with Chapter 2 of "Learning Deep Architectures for AI" at 3 pm in Ketcham Auditorium W1005.

It is a great opportunity for everyone to address what you understand about this topic or to ask questions about what you would like to learn. So I want to encourage everyone of us to attend. If you have friends who are interested in the area and want to join, please feel free to invite them. You can also contact me or Dr. DeSouza to add him/her on the list.

Hi MU CISers,

Tomorrow, we will have our first paper discussion on "Learning Deep Architectures for AI" at 3 pm in Ketcham Auditorium W1005. Likely we will start with the first couple of chapters. Depends on people's experience with Deep Learning area, we will decide to change the pace of the reading either faster or slower.

Here is the link to the dropbox folder:
https://www.dropbox.com/sh/lbvuntobvpv3hml/AACUXlfF85HXaJQp_I7jCBgaa?dl=0

To promote more group collaboration during the seminar series, in stead of doing a weekly presentation, we will start a biweekly group discussion on papers assigned by faculty members. For example, this week is still going to be the presentation but next week will be the group discussion. We will assign the paper for discussion a week before (probably on Tuesday) to give everyone enough time to read and study. Let me know if you have any questions regarding to this new approach of the CIS seminar!

Tomorrow, Jiao Changzhe will present a seminar on "Weakly Supervised Large Scale Object Localization with Multiple Instance Learning and Bag Splitting" at 3:00PM in Naka 120. (This is the last time using Naka 120. For the rest of this semester, all the seminar series will be held at Ketcham Auditorium W1005)

Abstract of the paper:
Localizing objects of interest in images when provided with only image-level labels is a challenging visual recognition task. Previous efforts have required carefully designed features and have difficulty in handling images with cluttered backgrounds. Up-scaling to large datasets also poses a challenge to applying these methods to real applications. In this paper, we propose an efficient and effective learning framework called MILinear, which is able to learn an object localization model from large-scale data without using bounding box annotations. We integrate rich general prior knowledge into a learning model using a large pre-trained convolutional network. Moreover, to reduce ambiguity in positive images, we present a bag-splitting algorithm that iteratively generates new negative bags from positive ones. We evaluate the proposed approach on the challenging Pascal VOC 2007 dataset, and our method outperforms other state-of-the-art methods by a large margin; some results are even comparable to fully supervised models trained with bounding box annotations. To further demonstrate scalability, we also present detection results on the ILSVRC 2013 detection dataset, and our method outperforms supervised deformable part-based model without using box annotations.

Hello MU CISers,

Welcome to the 2017 Spring semester!

Today Dr. Juan Gilbert is going to present "The Future of Elections Technology in the U.S.A." at 3 - 4pm in Ketcham Auditorium, W1005 Lafferre Hall.

Hello MU CISers,

Today Akshay Jain will present a seminar on "Modeling patterns of activities using activity curves" at 4:00 PM in EBW 353.

Abstract of the paper:
Pervasive computing offers an unprecedented opportunity to unobtrusively monitor behavior and use the large amount of collected data to perform analysis of activity-based behavioral patterns. In this paper, we introduce the notion of an activity curve, which represents an abstraction of an individual’s normal daily routine based on automatically-recognized activities. We propose methods to detect changes in behavioral routines by comparing activity curves and use these changes to analyze the possibility of changes in cognitive or physical health. We demonstrate our model and evaluate our change detection approach using a longitudinal smart home sensor dataset collected from 18 smart homes with older adult residents. Finally, we demonstrate how big data-based pervasive analytics such as activity curve-based change detection can be used to perform functional health assessment. Our evaluation indicates that correlations do exist between behavior and health changes and that these changes can be automatically detected using smart homes, machine learning, and big data-based pervasive analytics.

Hello MU CISers,



Tomorrow Xiaoxiao Du will present a seminar on "Adaptation and Evaluation of an Optical Flow Method Applied to Coregistration of Forest Remote Sensing Images" at 4:00 PM in EBW 353.



Abstract of the paper:

The coregistration of heterogeneous geospatial images is useful in various remote sensing applications. Since the number of available data increases and the resolution improves, it is interesting to have an approach as automated, fast, robust, and accurate as possible. In this paper, we present a solution based on optical-flow computation. This algorithm calledGeFolki allows the registration of images in a nonparametric and dense way. GeFolki is based on a local method of optical flow derived from the Lucas–Kanade algorithm, with a multiscale implementation, and a specific filtering including rank filtering, rolling guidance filtering and local contrast inversion. The efficiency of our coregistration chain is shown on radar, LIDAR, and optical images on Remningstorp forest in Sweden. An analysis of the relevant parameters is investigated for several scenarios. Finally,we demonstrate the accuracy of our coregistration by proposing specific metrics for LIDAR/radar coregistration, and optics/radar coregistration.

Hello MU CISers,

Due to a schedule conflict. The presentation from Xiaoxiao Du will be postponed to next Tuesday. Sorry for the inconvenience!

Hello MU CISers,

Tomorrow Xiaoxiao Du will present a seminar on "Adaptation and Evaluation of an Optical Flow Method Applied to Coregistration of Forest Remote Sensing Images" at 4:00 PM in EBW 353.

Abstract of the paper:
The coregistration of heterogeneous geospatial images is useful in various remote sensing applications. Since the number of available data increases and the resolution improves, it is interesting to have an approach as automated, fast, robust, and accurate as possible. In this paper, we present a solution based on optical-flow computation. This algorithm calledGeFolki allows the registration of images in a nonparametric and dense way. GeFolki is based on a local method of optical flow derived from the Lucas–Kanade algorithm, with a multiscale implementation, and a specific filtering including rank filtering, rolling guidance filtering and local contrast inversion. The efficiency of our coregistration chain is shown on radar, LIDAR, and optical images on Remningstorp forest in Sweden. An analysis of the relevant parameters is investigated for several scenarios. Finally,we demonstrate the accuracy of our coregistration by proposing specific metrics for LIDAR/radar coregistration, and optics/radar coregistration.

Hello MU CISers,

Today Plodpradista Pooparat will present a seminar on "Log-Gabor Filters for Image-Based Vehicle Verification" at 4:00 PM in EBW 353.

Abstract of the paper:
Vehicle detection based on image analysis has attracted increasing attention in recent years due to its low cost, flexibility, and potential toward collision avoidance. In particular, vehicle verification is especially challenging on account of the heterogeneity of vehicles in color, size, pose, etc. Imagebased vehicle verification is usually addressed as a supervised classification problem. Specifically, descriptors using Gabor filters have been reported to show good performance in this task.
However, Gabor functions have a number of drawbacks relating to their frequency response. The main contribution of this paper is the proposal and evaluation of a new descriptor based on the alternative family of log-Gabor functions for vehicle verification, as opposed to existing Gabor filter-based descriptors. These filters are theoretically superior to Gabor filters as they can better represent the frequency properties of natural images. As a second contribution, and in contrast to existing approaches, which transfer the standard configuration of filters used for other applications to the vehicle classification task, an in-depth analysis of the required filter configuration by both Gabor and log-Gabor descriptors for this particular application is performed for fair comparison. The extensive experiments conducted in this paper confirm that the proposed log-Gabor descriptor significantly outperforms the standard Gabor filter for image-based vehicle verification.