Browsing by Subject "classification"
Now showing 1 - 4 of 4
Results Per Page
Sort Options
Item A Compressive Sensing Approach To Hyperspectral Image Classification(2019-01-01) Della Porta, Charles John; Chang, Chein-I; Computer Science and Electrical Engineering; Engineering, ElectricalHyperspectral imaging (HSI) technology has found success in a variety of applications; however, its use is often still limited due to size, weight and power (SWaP) constraints. In this dissertations, compressive sensing (CS) is proposed as an enabling technology to reduce the high spectral band count, through the creation of compressively-sensed bands (CSBs). A CS model based on the universality of random sensing is proposed for the analysis of hyperspectral classification in the compressed domain. Specifically, the utility of the support vector machine (SVM) in the compressed domain is evaluated through both mathematical analysis and empirical experimentation. This work shows that is indeed possible to achieve full band classification performance in the compressed domain. The experiments also demonstrate that the minimum number of CSBs is scene dependent, requiring additional algorithms to provide a full solution. Two supervised algorithms based on a feature selection framework are proposed for estimating the minimum lower bound on the required number of CSBs. The first algorithm is based on feature filtering techniques and the second algorithm is based on classifier wrapping. Finally, an unsupervised algorithm is presented, based on progressive band processing, that is able to adaptively determine the required number of CSBs in-situ. The contributions of this dissertations provide the fundamental foundation for compressed classification of hyperspectral images and identify several new opportunities for future research.Item Detecting Makeup Activities using Internet-of-Things(2019-01-01) Alqurmti, Fatimah; Roy, Nirmalya; Information Systems; Information SystemsThis theses focuses on identifying human activities for rendering make-up activities using sensors' data and a supervised machine learning approaches. We considered five make-up activities in our work, such as, applying cream, lipsticks, blusher, eyeshadow, and mascara. We collected the data from ten participants using two smart-watch built-in sensors, accelerometer and gyroscope. We preprocessed the data and trained with different predictive machine learning models and we evaluated make-up activity prediction built on using Naïve Bayes, Simple Logistic, k-nearest neighbors', and the random forest algorithms. We investigated the models' performance on three different datasets that differ by the environment they were collected in. The first dataset was collected from the participants using a controlled environment. In this staged setting, we provided the participants specific instructions on how to perform the five make-up activities. The second dataset was collected from the participants in an uncontrolled environment. We did not inform the participants with any prior instructions on how to perform the five activities and therefore, naturally they performed the make-up activities in their own way. Third, we synthetically generated a dataset by combining the existing datasets from the participants who were under both controlled and uncontrolled environments. Our results showed a 92.7 % accuracy for the controlled environment case given by the Gradient Boosting classifier and an 89.20 % accuracy for the uncontrolled environment case shown by the Random Forest classifier. Finally, Random Forest classifier registered the highest accuracy 92%, for the hybrid case where both the datasets from controlled and the uncontrolled environments were combined. We believe that this early work on recognizing and discovering a multitude of make-up activities has potential application in assessing and training the performance of various stakeholders in the future work of fashion industry.Item Differential Fairness for Machine Learning and Artificial Intelligence Systems: Unbiased Decisions with Biased Data(2018-11-14) Foulds, James; Islam, Rashidul; Keya, Kamrun; Pan, ShimeiItem Weakly Supervised Cascaded Convolutional Networks(IEEE, 2017-07-26) Diba, Ali; Sharma, Vivek; Pazandeh, Ali; Pirsiavash, Hamed; Gool, Luc VanObject detection is a challenging task in visual understanding domain, and even more so if the supervision is to be weak. Recently, few efforts to handle the task without expensive human annotations is established by promising deep neural network. A new architecture of cascaded networks is proposed to learn a convolutional neural network (CNN) under such conditions. We introduce two such architectures, with either two cascade stages or three which are trained in an end-to-end pipeline. The first stage of both architectures extracts best candidate of class specific region proposals by training a fully convolutional network. In the case of the three stage architecture, the middle stage provides object segmentation, using the output of the activation maps of first stage. The final stage of both architectures is a part of a convolutional neural network that performs multiple instance learning on proposals extracted in the previous stage(s). Our experiments on the PASCAL VOC 2007, 2010, 2012 and large scale object datasets, ILSVRC 2013, 2014 datasets show improvements in the areas of weakly-supervised object detection, classification and localization.