Hearing aids aim to process and modify sounds into the most desirable forms for hearing impaired people to receive. However, due to multiple reasons including inconvenience and limited quality, only 20 percent of the people in the US who could benefit from a hearing aid wear one. This figure is likely to be much lower in other less developed countries.
Recently, smartphones with powerful computation capability and great mobility have emerged as a possible alternative for this problem. We have developed a preliminary iOS application with certain sound processing functionalities. It is able to collect all the sounds in the vicinity and amplify custom frequencies depending on the prescriptions of a specific user. In addition, the application can also produce different output on either the left or the right headphone piece. We have taken initial steps to make the system operate wirelessly with a Bluetooth earpiece; however, due to time and resources constraints, the application has not yet able to divide two distinct output like what it does on the normal iPhone earpiece. Also, a method for shifting sounds to lower frequency has not yet been implemented. We also have not yet tested the program to its fullest potential due to the sole access to only built-in iPhone’s microphone. A special microphone with many features such as noises canceling, separate streaming, and high sampling rate will enables us fully customize and prepare the application for future technologies. Our future system is expected to address these challenges.
Recent advances in image recognition have been catalyzed by progressions in the applications of convolutional neural networks (CNN) and deep learning (DL). In traditional artificial intelligence (AI), neural networks (NN) were represented in a “shallow” fashion; dictating only one dimensional vectors at various layers. Furthermore past networks were often confined to three main layers: input, hidden and output layers. This rigidity of the structure not only contrasted the NN’s derivation from complicated biological neural systems but also limited their capability of categorizing inputs of various sizes and orientations (like images.) CNN's sought to solve this problem by representing a NN in terms of 3D volumes in which a kernel is moved in a sliding manner over subsections of an input volume and convolved with these regions to generate a k-layer output volume. This output volume is comprised of filtered versions of the previous volume which help detect recognizable features while maintaining important spatial features. This project created a deep CNN which leverages the Java library DeepLearning4j to demonstrate these techniques and provide a simple program which attempts to categorize input images into one of 5 classes.
In U.S., about 63% of households include pets. However, certain pets (such as dogs) have the instinct to run away from the house. Yet, it is impossible for the pet owners to watch their pets all the time. Therefore, a portable and inexpensive handheld tracking system can be a useful tool for helping the owners to watch their pets.
This project intend to employ iBeacon, which is a technology released by Apple Inc., to build a tracking system. The iBeacon technique can achieve distance measurements based on the Received Signal Strength (RSS). The RSS value will change as the distance between Beacon and the signal receiving device change. Moreover, the iBeacon tag device for pets (called iBeacon tags) is small (in the size of a quarter) enough to put on the collar of a pet. The application will store the information of beacons (including UUID, which is used to distinguish different beacons) that provide by users, and continually detect the signal from the beacons. When the signal is not strong enough, which means the Beacon is out of the controllable range, then the application will alert the user.
Author(s): Kathryn Jaslikowski Computer Science Nick Bomm Computer Science Phil Howell Computer Science Wills Ward Computer Science
Advisor(s): Donnell Payne Computer Science
Location: Session: 2; B0; Table Number: 5
The purpose of this capstone project is to aid Meals on Wheels (MOW) of Tarrant County with in-home consultations for low-income and low-mobility clients. MOW can only provide its clients with 10 meals per week and so, in conjunction with the TCU Nutrition Department, approached our group about creating a website to provide clients with easy-to-make, low-cost recipes to supplement for the week’s other meals. We then developed a website that allows dietitians to develop recipe and shopping lists based on client food preferences, allergies, and appliance/utensil restrictions. Dietitians can then print a PDF file of the recipes and shopping lists in-home for clients to keep. Foods, recipes, and stores can be dynamically added, edited, and deleted from the database by administrators and interns. We also calculate the nutritional information for each recipe using a USDA Nutrient Database API to ensure that the MOW clients are fulfilling their nutritional needs.
Author(s): Rebecca Ruch Computer Science Cameron Diou Computer Science Harrison Engel Computer Science Steven Garcia Computer Science Will Taylor Computer Science
Advisor(s): Donnell Payne Computer Science Lisa Ball Computer Science
Location: Session: 1; 2nd Floor; Table Number: 7
Expanding Your Horizons Network (EYHN) is a 501(c)3 nonprofit organization dedicated to providing gateway STEM (Science, Technology, Engineering, and Math) experiences to middle and high school girls that spark interest in activities and careers within these fields. EYHN accomplishes this through role-model led conferences with hands on STEM activities and workshops.
These conferences are hosted by various organizations across the country. In Fort Worth, an annual EYHN conference is hosted by Texas Wesleyan University (TxWes). Each year, this conference hosts hundreds of student participants and requires dozens of leaders, volunteers, and presenters. Handling a conference of this size requires significant organizational effort, with a bulk of pre-conference administrative work going to registering participants and creating a good schedule for the event. In previous years, organizers at TxWes used a scheduling and registration system created by TCU students in 2005. However, this program is out of date and no longer useable making a replacement necessary.
This Project, Scheduling Your Horizons (SYH), shall create a replacement system for TxWes that allows TxWes organizers to register participants and generate a schedule for the conference. It aims to do so in a modern, user-friendly manner, with an emphasis on platform independence and maintainability to extend the lifespan of the application.
Author(s): James Stewart Computer Science Michael Giba Computer Science Quang Nguyen Computer Science Son Nguyen Computer Science Thaddeus Rix Computer Science
Advisor(s): Donnell Payne Computer Science Billy Farmer Computer Science Liran Ma Computer Science
Location: Session: 2; 3rd Floor; Table Number: 9
<b>TCU’s previous Student Research Symposium site provided an outdated submission-review system for the Michael and Sally McCracken Student Research Symposium, an event growing in popularity. The old system was mostly a front-end to a primarily manual collection of procedures to collect, review, and present research projects. There was a growing need to make a more robust system that can provide smart interfaces for various users that allows for secure submitting, balloting, and administration.</b>
The new system provides a host of automated processes that facilitate in the management of the SRS event from year to year, including such things as automatic archiving of previous year’s information. This is possible due to a myriad of free technologies such as Django. To complement the many processes we have automated, we have created tools for administrators to change information in the website without entering the codebase. Among the automated processes and features that help with administration, we have embedded advanced algorithms which reduce the need for human involvement, such as cost-analysis table assignments, a procedure that once required hours of laborious calculations.