Fast cross-validation of kernel fisher discriminant classifiers

Given n training examples, the training of a Kernel Fisher Discriminant (KFD) classifier corresponds to solving a linear system of dimension n. In cross-validating KFD, the training examples are split into 2 distinct subsets for a number of times (L) wherein a subset of m examples is used for validation and the other subset of(n - m) examples is used for training the classifier. In this case L linear systems of dimension (n - m) need to be solved. We propose a novel method for cross-validation of KFD in which instead of solving L linear systems of dimension (n - m), we compute the inverse of an n × n matrix and solve L linear systems of dimension 2m, thereby reducing the complexity when L is large and/or m is small. For typical 10-fold and leave-one-out cross-validations, the proposed algorithm is approximately 4 and (^4/9n) times respectively as efficient as the naive implementations. Simulations are provided to demonstrate the efficiency of the proposed algorithms.

Image source detection : a case study on Facebook images taken by iPhones

Forged and tempered digital images become increasingly common on Facebook to aid computer frauds. The situation is worsened as many users can use a phone to take a photo and upload it to Facebook within two clicks, which highlights the need of image forensics for the cyber fraud cases. In this paper, we show the existence of the Facebook image filter which automatically changes the Facebook photos and consequently challenges the validity of forensic results. We aim to enable forensic investigators to relate a seized camera and a Facebook image. Specifically, we utilize intrinsic sensor pattern noise produced by a camera's lens to derive forensically useful information as Photo Response Non-Uniformity (PRNU) patterns. We propose to compare the PRNU patterns of a Facebook image and the flat field images produced by the candidate cameras. And we conclude this method to be effective by successfully identifying the correct iPhone from a list of four for a given Face book image.

Computer-aided detection of depression from magnetic resonance images

Magnetic resonance imaging (MRI) of the brain is used to detect depression disorder. However, a large number of MRI scans needs to be analyzed for such detection. Manual segmentation of the biomarkers in MRI scans by clinical experts can become time consuming and sometimes erroneous. This paper presents a study on computer-aided detection of depression from MRI scans. These systems have not yet been identified, categorized and compared in the literature. The paper covers fully automated to semi-automated detection systems. It also presents performance comparison for the considered systems.

Application of growing self-organizing map to distinguish between finger tapping and non tapping from brain images

Growing self-organizing map (GSOM) has been characterized as a knowledge discovery visualization application which outshines the traditional self-organizing map (SOM) due to its dynamic structure in which nodes can grow based on the input data. GSOM is utilized as a visualization tool in this paper to cluster fMRI finger tapping and non- tapping data, demonstrating the visualization capability to distinguish between tapping or non-tapping. A unique feature of GSOM is a parameter called the spread factor whose functionality is to control the spread of the GSOM map. By setting different levels of spread factor, different granularities of region of interests within tapping or non-tapping images can be visualized and analyzed. Euclidean distance based similarity calculation is used to quantify the visualized difference between tapping and non tapping images. Once the differences are identified, the spread factor is used to generate a more detailed view of those regions to provide a better visualization of the brain regions.

Evaluation of segmentation algorithms for extraction of RNFL in OCT images

The thickness of the retinal nerve fiber layer (RFNL) has become a diagnose measure for glaucoma assessment. To measure this thickness, accurate segmentation of the RFNL in optical coherence tomography (OCT) images is essential. Identification of a suitable segmentation algorithm will facilitate the enhancement of the RNFL thickness measurement accuracy. This paper investigates the performance of six algorithms in the segmentation of RNFL in OCT images. The algorithms are: normalised cuts, region growing, k-means clustering, active contour, level sets segmentation: Piecewise Gaussian Method (PGM) and Kernelized Method (KM). The performance of the six algorithms are determined through a set of experiments on OCT retinal images. An experimental procedure is used to measure the performance of the tested algorithms. The measured segmentation precision-recall results of the six algorithms are compared and discussed.