Matching Items (31)

Learning Users Visual Preferences: Building a Recommendation System for Instagram

Description

Social media users are inundated with information. Especially on Instagram--a social media service based on sharing photos--where for many users, missing important posts is a common issue. By creating a

Social media users are inundated with information. Especially on Instagram--a social media service based on sharing photos--where for many users, missing important posts is a common issue. By creating a recommendation system which learns each user's preference and gives them a curated list of posts, the information overload issue can be mediated in order to enhance the user experience for Instagram users. This paper explores methods for creating such a recommendation system. The proposed method employs a learning model called ``Factorization Machines" which combines the advantages of linear models and latent factor models. In this work I derived features from Instagram post data, including the image, social data about the post, and information about the user who created the post. I also collect user-post interaction data describing which users ``liked" which posts, and this was used in models leveraging latent factors. The proposed model successfully improves the rate of interesting content seen by the user by anywhere from 2 to 12 times.

Contributors

Agent

Created

Date Created
  • 2016-12

135858-Thumbnail Image.png

A Statistical Framework for Detecting Edges from Noisy Fourier Data with Multiple Concentration Factors

Description

The concentration factor edge detection method was developed to compute the locations and values of jump discontinuities in a piecewise-analytic function from its first few Fourier series coecients. The method

The concentration factor edge detection method was developed to compute the locations and values of jump discontinuities in a piecewise-analytic function from its first few Fourier series coecients. The method approximates the singular support of a piecewise smooth function using an altered Fourier conjugate partial sum. The accuracy and characteristic features of the resulting jump function approximation depends on these lters, known as concentration factors. Recent research showed that that these concentration factors could be designed using aexible iterative framework, improving upon the overall accuracy and robustness of the method, especially in the case where some Fourier data are untrustworthy or altogether missing. Hypothesis testing methods were used to determine how well the original concentration factor method could locate edges using noisy Fourier data. This thesis combines the iterative design aspect of concentration factor design and hypothesis testing by presenting a new algorithm that incorporates multiple concentration factors into one statistical test, which proves more ective at determining jump discontinuities than the previous HT methods. This thesis also examines how the quantity and location of Fourier data act the accuracy of HT methods. Numerical examples are provided.

Contributors

Agent

Created

Date Created
  • 2016-05

135973-Thumbnail Image.png

An l1 Regularization Algorithm for Reconstructing Piecewise Smooth Functions from Fourier Data Using Wavelet Projection

Description

Imaging technologies such as Magnetic Resonance Imaging (MRI) and Synthetic Aperture Radar (SAR) collect Fourier data and then process the data to form images. Because images are piecewise smooth, the

Imaging technologies such as Magnetic Resonance Imaging (MRI) and Synthetic Aperture Radar (SAR) collect Fourier data and then process the data to form images. Because images are piecewise smooth, the Fourier partial sum (i.e. direct inversion of the Fourier data) yields a poor approximation, with spurious oscillations forming at the interior edges of the image and reduced accuracy overall. This is the well known Gibbs phenomenon and many attempts have been made to rectify its effects. Previous algorithms exploited the sparsity of edges in the underlying image as a constraint with which to optimize for a solution with reduced spurious oscillations. While the sparsity enforcing algorithms are fairly effective, they are sensitive to several issues, including undersampling and noise. Because of the piecewise nature of the underlying image, we theorize that projecting the solution onto the wavelet basis would increase the overall accuracy. Thus in this investigation we develop an algorithm that continues to exploit the sparsity of edges in the underlying image while also seeking to represent the solution using the wavelet rather than Fourier basis. Our method successfully decreases the effect of the Gibbs phenomenon and provides a good approximation for the underlying image. The primary advantages of our method is its robustness to undersampling and perturbations in the optimization parameters.

Contributors

Agent

Created

Date Created
  • 2015-12

147616-Thumbnail Image.png

The Role of Fourier Phase in Image Representation and Reconstruction

Description

The Fourier representation of a signal or image is equivalent to its native representation in the sense that the signal or image can be reconstructed exactly from its Fourier transform.

The Fourier representation of a signal or image is equivalent to its native representation in the sense that the signal or image can be reconstructed exactly from its Fourier transform. The Fourier transform is generally complex-valued, and each value of the Fourier spectrum thus possesses both magnitude and phase. Degradation of signals and images when Fourier phase information is lost or corrupted has been studied extensively in the signal processing research literature, as has reconstruction of signals and images using only Fourier magnitude information. This thesis focuses on the case of images, where it examines the visual effect of quantifiable levels of Fourier phase loss and, in particular, studies the merits of introducing varying degrees of phase information in a classical iterative algorithm for reconstructing an image from its Fourier magnitude.

Contributors

Agent

Created

Date Created
  • 2021-05

133580-Thumbnail Image.png

Image Processing for an Autonomous Throwing Arm and Smart Catching System

Description

In this paper, we propose an autonomous throwing and catching system to be developed as a preliminary step towards the refinement of a robotic arm capable of improving strength and

In this paper, we propose an autonomous throwing and catching system to be developed as a preliminary step towards the refinement of a robotic arm capable of improving strength and motor function in the limb. This will be accomplished by first autonomizing simpler movements, such as throwing a ball. In this system, an autonomous thrower will detect a desired target through the use of image processing. The launch angle and direction necessary to hit the target will then be calculated, followed by the launching of the ball. The smart catcher will then detect the ball as it is in the air, calculate its expected landing location based on its initial trajectory, and adjust its position so that the ball lands in the center of the target. The thrower will then proceed to compare the actual landing position with the position where it expected the ball to land, and adjust its calculations accordingly for the next throw. By utilizing this method of feedback, the throwing arm will be able to automatically correct itself. This means that the thrower will ideally be able to hit the target exactly in the center within a few throws, regardless of any additional uncertainty in the system. This project will focus of the controller and image processing components necessary for the autonomous throwing arm to be able to detect the position of the target at which it will be aiming, and for the smart catcher to be able to detect the position of the projectile and estimate its final landing position by tracking its current trajectory.

Contributors

Agent

Created

Date Created
  • 2018-05

133444-Thumbnail Image.png

Analyzing rat sciatic nerve fibers under various electrical stimuli

Description

Abstract Modern imaging techniques for sciatic nerves often use imaging techniques that can clearly find myelinated axons (Group A and Group B and analyze their properties, but have trouble with

Abstract Modern imaging techniques for sciatic nerves often use imaging techniques that can clearly find myelinated axons (Group A and Group B and analyze their properties, but have trouble with the more numerous Remak Fibers (Group C). In this paper, Group A and B fibers are analyzed while also analyzing Remak fibers using osmium tetroxide staining and imaging with the help of transmission electron microscopy. Using this method, nerves had various electrical stimuli attached to them and were analyzed as such. They were analyzed with a cuff electrode attached, a stimulator attached, and both, with images taken at the center of the nerve and the ends of them. The number and area taken by the Remak fibers were analyzed, along with the g-ratios of the Group A and B fibers. These were analyzed to help deduce the overall health of the fibers along with vacuolization, and mitochondria available. While some important information was gained from this evaluation, further testing has to be done to improve the myelin detection system, along with analyzing the proper and necessary Remak fibers and the role they play. The research tries to thoroughly look at the necessary material and find a way to use it as a guide to further experimentation with electrical stimuli, and notes the differences found within and without various groups, various points of observation, and various stimuli as a whole. Nevertheless, this research allows a strong look into the benefits of transmission electron microscopy and the ability to assess electrical stimulation from these points.

Contributors

Created

Date Created
  • 2018-05

148467-Thumbnail Image.png

Application of See-Through Car Pillars in the Automobile Industry

Description

This creative project is an extension of the work being done as part of Senior Design in<br/>developing the See-Through Car Pillar, a system designed to render the forward car pillars

This creative project is an extension of the work being done as part of Senior Design in<br/>developing the See-Through Car Pillar, a system designed to render the forward car pillars in a car<br/>invisible to the driver so they can have an unobstructed view utilizing displays, sensors, and a<br/>computer. The first half of the paper provides the motivation, design and progress of the project, <br/>while the latter half provides a literature survey on current automobile trends, the viability of the<br/>See-Through Car Pillar as a product in the market through case studies, and alternative designs and <br/>technologies that also might address the problem statement.

Contributors

Agent

Created

Date Created
  • 2021-05

152360-Thumbnail Image.png

Image processing using approximate data-path units

Description

In this work, we present approximate adders and multipliers to reduce data-path complexity of specialized hardware for various image processing systems. These approximate circuits have a lower area, latency and

In this work, we present approximate adders and multipliers to reduce data-path complexity of specialized hardware for various image processing systems. These approximate circuits have a lower area, latency and power consumption compared to their accurate counterparts and produce fairly accurate results. We build upon the work on approximate adders and multipliers presented in [23] and [24]. First, we show how choice of algorithm and parallel adder design can be used to implement 2D Discrete Cosine Transform (DCT) algorithm with good performance but low area. Our implementation of the 2D DCT has comparable PSNR performance with respect to the algorithm presented in [23] with ~35-50% reduction in area. Next, we use the approximate 2x2 multiplier presented in [24] to implement parallel approximate multipliers. We demonstrate that if some of the 2x2 multipliers in the design of the parallel multiplier are accurate, the accuracy of the multiplier improves significantly, especially when two large numbers are multiplied. We choose Gaussian FIR Filter and Fast Fourier Transform (FFT) algorithms to illustrate the efficacy of our proposed approximate multiplier. We show that application of the proposed approximate multiplier improves the PSNR performance of 32x32 FFT implementation by 4.7 dB compared to the implementation using the approximate multiplier described in [24]. We also implement a state-of-the-art image enlargement algorithm, namely Segment Adaptive Gradient Angle (SAGA) [29], in hardware. The algorithm is mapped to pipelined hardware blocks and we synthesized the design using 90 nm technology. We show that a 64x64 image can be processed in 496.48 µs when clocked at 100 MHz. The average PSNR performance of our implementation using accurate parallel adders and multipliers is 31.33 dB and that using approximate parallel adders and multipliers is 30.86 dB, when evaluated against the original image. The PSNR performance of both designs is comparable to the performance of the double precision floating point MATLAB implementation of the algorithm.

Contributors

Agent

Created

Date Created
  • 2013

154269-Thumbnail Image.png

Effective gene expression annotation approaches for mouse brain images

Description

Understanding the complexity of temporal and spatial characteristics of gene expression over brain development is one of the crucial research topics in neuroscience. An accurate description of the locations and

Understanding the complexity of temporal and spatial characteristics of gene expression over brain development is one of the crucial research topics in neuroscience. An accurate description of the locations and expression status of relative genes requires extensive experiment resources. The Allen Developing Mouse Brain Atlas provides a large number of in situ hybridization (ISH) images of gene expression over seven different mouse brain developmental stages. Studying mouse brain models helps us understand the gene expressions in human brains. This atlas collects about thousands of genes and now they are manually annotated by biologists. Due to the high labor cost of manual annotation, investigating an efficient approach to perform automated gene expression annotation on mouse brain images becomes necessary. In this thesis, a novel efficient approach based on machine learning framework is proposed. Features are extracted from raw brain images, and both binary classification and multi-class classification models are built with some supervised learning methods. To generate features, one of the most adopted methods in current research effort is to apply the bag-of-words (BoW) algorithm. However, both the efficiency and the accuracy of BoW are not outstanding when dealing with large-scale data. Thus, an augmented sparse coding method, which is called Stochastic Coordinate Coding, is adopted to generate high-level features in this thesis. In addition, a new multi-label classification model is proposed in this thesis. Label hierarchy is built based on the given brain ontology structure. Experiments have been conducted on the atlas and the results show that this approach is efficient and classifies the images with a relatively higher accuracy.

Contributors

Agent

Created

Date Created
  • 2016

155174-Thumbnail Image.png

Monitoring physiological signals using camera

Description

Monitoring vital physiological signals, such as heart rate, blood pressure and breathing pattern, are basic requirements in the diagnosis and management of various diseases. Traditionally, these signals are measured only

Monitoring vital physiological signals, such as heart rate, blood pressure and breathing pattern, are basic requirements in the diagnosis and management of various diseases. Traditionally, these signals are measured only in hospital and clinical settings. An important recent trend is the development of portable devices for tracking these physiological signals non-invasively by using optical methods. These portable devices, when combined with cell phones, tablets or other mobile devices, provide a new opportunity for everyone to monitor one’s vital signs out of clinic.

This thesis work develops camera-based systems and algorithms to monitor several physiological waveforms and parameters, without having to bring the sensors in contact with a subject. Based on skin color change, photoplethysmogram (PPG) waveform is recorded, from which heart rate and pulse transit time are obtained. Using a dual-wavelength illumination and triggered camera control system, blood oxygen saturation level is captured. By monitoring shoulder movement using differential imaging processing method, respiratory information is acquired, including breathing rate and breathing volume. Ballistocardiogram (BCG) is obtained based on facial feature detection and motion tracking. Blood pressure is further calculated from simultaneously recorded PPG and BCG, based on the time difference between these two waveforms.

The developed methods have been validated by comparisons against reference devices and through pilot studies. All of the aforementioned measurements are conducted without any physical contact between sensors and subjects. The work presented herein provides alternative solutions to track one’s health and wellness under normal living condition.

Contributors

Agent

Created

Date Created
  • 2016