CS Night 2018

Lo Schiavo Hall and Walkway

Join the Department of Computer Science for the 17th Annual Computer Science Night 

Thursday, Dec. 6 |  5:30–8:30 p.m.
John Lo Schiavo, S.J. Center for Science and Innovation | USF Campus


Reserve your spot »

This year's research and project abstracts

 

Java Profiler & Visualization Tool
Presenter: Brian Bushree
Faculty Advisors: Sami Rollins, Greg Benson

A Java profiling tool that allows a user to profile and visualize a Java program’s execution and display each of its read/write instructions. This program uses a technology known as bytecode injection to record pieces of the program’s state at runtime. This tool is targeted at both beginner developers searching for bugs and advanced developers trying to visualize and conceptualize large-scale Java programs. Though in a primitive state, this application has potential to be expanded upon and serve as a developing tool to run alongside an IDE to better conceptualize and visualize Java programs.

 

Pipeline Orchestration For SnapLogic
Presenters: Cristina Chu, Kundry Rivero, Jovani Rico
Sponsor: Anju Schiller, SnapLogic
Faculty Advisor: CS 690 Master's Project. Prof. Olga Karpenko

Currently, SnapLogic’s customers can only view data flow pipelines on their own, with no relation to other pipelines. This absence of an overarching view can affect tracking data movement across pipelines and data locations, which can lead to problems such as having duplicate pipelines doing the same work and data sources underperforming due to excessive traffic. The goal of this project is to provide more information about the relationship between pipelines and data locations through a visual tool. Using the tool, a customer will be able to see all data locations used by their organization and see the interacting pipelines during a period of time.

 

Fitness Tracker
Presenters: Gudbrand Schistad, Hassan Chadad, Lovedeep Sandhu, Omar Sharif
Sponsor: Jose Alvarado
Faculty Advisor: CS 690 Master's Project. Prof. Olga Karpenko

A common problem for people who regularly exercise is the difficulty of tracking fitness progress. Our mobile application (available for both Android and iOS devices) provides a platform to organize tasks, keep track of time, and log activities for workouts with minimal effort. In addition to these core features, our application has more advanced features such as the ability to be awarded badges to mark fitness progress, ability to follow other user profiles, view progress analysis charts, search gyms, and calculate expenses. These features as well as other features of our mobile app assist a user in tracking workouts, strength gains, body measurements, calories burned, and food consumption. Users of Fitness Tracker will have an easier time tracking workout progress, staying motivated, organizing their tasks, and learning more about fitness.

 

Understanding and Measuring Ethereum
Presenters: Charles (Yunjie) Ge, Jinyue Song
Sponsor: Prof. Eunjin (EJ) Jung
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

As an open software platform, Ethereum enables developers to build and deploy decentralized applications without worrying about implementing the blockchain technology. All of the exchange of Ether coins or transferring of data are represented as transactions in Ethereum. The first goal of our project is to find the relationship between the transaction fee and the time for the transaction to be confirmed on the canonical chain. Our code automates the process of data cleaning, storage, and analysis. Sometimes, more than one block might be created at the same time, which would cause a fork. The second goal of our project is to track high-way forks. It is hard for Ethereum to use Proof of Stake instead of Proof of Work to create new blocks if high-way fork happens. Similar to the real financial environment, Ponzi schemes exist in Ethereum. We added more features to Professor EJ Jung's previous Ponzi scheme detection models to help improve their precision.

 

Ethereum Intelligence
Presenters: Brian Sung, Nate Wilson,
Sponsor: Jim Huang
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

As blockchain technologies such as Ethereum become further used and more invested in, it is essential to have infrastructure in place for doing data analysis so as to analyze the performance of these networks. In this project we set up a full Ethereum node, and configure it to be used for high volume data analysis of the blockchain. We use Ethereum-ETL to pull the data from our node and pandas and jupyter notebooks to carry out data analysis.

To prove the value of our analysis infrastructure, we present three selected analyses of various aspects of the Ethereum blockchain. Our first exploration takes a detailed look at block rewards and how they affect the motivation of miners. Next we present a method for visualizing soft forks and the uncle blocks they create. Finally we present a simple logistic regression model for detecting the presence of private mining pools.

 

Finding Strong Gravitational Lenses with Residual Neural Networks
Matthew Domingo, Varun Ravi
Sponsor: Prof. Xiaosheng Huang
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

Measuring gravitational lensing by galaxies is the only way to directly study the elusive dark matter. However, gravitational lensing is a very rare phenomenon (~1 in 10,000 galaxies). Our goal is to find new strong gravitational lenses using deep neural networks (“neural nets”). We train our neural nets on a hand-labeled set of images, consisting of both lenses and non-lenses (“the training set”). We then apply the trained neural nets to a “validation set” to assess the accuracy and precision of its predictions.  Our data sets are selected from real observational data.

 

Microsavings
Presenters: Katherine Tucto, Diana Perez
Sponsor: Shadi Saifan, Trizic
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

Trizic is a company that provides digital wealth solutions to large financial institutions. The company has developed an interest in providing financial advice to individuals by building an end-user savings solution.

For this project, we developed Microsavings, a goal-based solution that provides recommendations to end-users by identifying trends in their bank accounts activity, enabling them to make small saving transactions from multiple accounts. The software will allow users to create an account where they will be able to enter their financial information and intended saving goals so that optimal recommendations can be generated by the algorithm. The saving goals are configurable parameters that can be modified at any time by the user.

 

AI for Gaia Project
Presenters: Yifan Zhou, Zheng Li
Sponsor: Jon Rahoi
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

In Gaia Project, each player controls one of 14 factions striving to peacefully colonize the Terra Mystica galaxy. During the game, the player will take actions to get as many victory points as possible, such as building mines, upgrading structures, doing research and forming federations. In the end, after six rounds of play, the faction with the most victory points wins. We explored three different approaches (Random Bot, Tree Searching, Neural Network) to build the AI  for this game.

 

Optimizing Yelp's Network Communication Library
Presenters: Albert Yang, Peilin Zhong, Ryan Yu, Timothy Yung
Sponsor: John Billings
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

Bravado is a Python library used for service to service communication. It aims to be a complete replacement for code generation, which can be validated through Swagger specification. It is a popular library that has been widely used in Yelp. Bravado-core is a library used to fetch, validate and unmarshal the Swagger specification for Bravado library. At present, there are some inefficiencies when passing large data sets which limits the number of engineers who can fully utilize this tool. In this project we explored many different ways to optimize code, ranging from optimizations based on the logic of the program to generic optimizations that can be used for any code written in Python.

 

FightMaker
Presenters: Sheng Chen, Kai Yu, Martino Kuan
Sponsor: Jon Rahoi
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

FightMaker is a web application to manage and plan fight events involving multiple fights.  Fights may be coordinated with the help of integration between the fighter database and the event planner user interface.  In addition, anyone can view the schedules of upcoming fights and details of their favorite fighters.  Coaches can manage the information regarding their schools and fighters, while promoters can prepare new events and fight matches.

 

FiTX - Tinder inspired platform for fitness
Presenters: Chirag Jain, Nikhil Barapatre, Prateek Tiwari
Sponsor: Jose Alvarado
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

Often, people who are dedicatedly following a fitness regime find it difficult to continue with the same enthusiasm. For instance, they can very easily get bored of doing the same thing everyday and eventually, lose motivation towards going to the gym. FiTX is the perfect solution for this problem as it matches users based on their fitness preferences. Thus, reviving the motivation to continue with their fitness goals. Our mobile application aims to provide a platform for fitness enthusiasts to find other people with similar fitness interests and goals. We believe that, having a fitness partner can play a substantial l role in keeping its users motivated to be on track in their fitness journey.


Crypto Tracker: Managing Crypto Portfolios
Presenters: Yuxiao Yang, Xin Cheng
Sponsor: Jose Alvarado
Faculty Advisor: CS 690 Master's Project, Prof. Olga Karpenko

Crypto tracker is cryptocurrency portfolio management platform integrated with CoinBase. It allows users to track their portfolio performance over time and computes the cost basis and capital gains to make it easier for the user to file taxes. Crypto tracker’s intended users include new investors and those who are familiar with cryptocurrencies. Those who are new to cryptocurrency investment can manage their crypto assets just as easily as managing their bank accounts. Professional crypto investors can utilize Crypto Tracker to get complete, updated and detailed information about the whole market, and take advantage of visualizations to gain more insight into investment strategies.


Investigating Implicit Gender Bias using Full Body Visuomotor Synchrony in Virtual Reality
Presenters: Sarah Lopez, Yi Yang, Kevin Beltran, Jennifer Cruz Hernandez, Chelsy Simran, Soojung Kim, Bingkun Yang
Sponsor: Professor Beste Yuksel, Human-Computer Interaction Lab, Department of Computer Science
Faculty Advisor: Professor Beste Yuksel

Previous research has shown that when white people embody a black avatar in virtual reality (VR) with full body visuomotor synchrony, this can reduce their implicit racial bias. We put men in female and male avatars in VR with full visuomotor synchrony using wearable trackers and investigated implicit gender bias and embodiment. We discuss the future implications of the significant findings for both VR scenarios and embodiment technologies. We'll be live demoing the full body tracking in VR at CS night. 
 


Recited Verse: Connecting people passionate about poetry
Presenters: Alice Zhang, Kuan Wang, QiaoJian Hu, Enxy Wu
Sponsor: Omar Miranda
Faculty Advisors: Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

Recited Verse is a social community dedicated to creating and sharing original audio recordings of poetry. In the spirit of Soundcloud and Spotify, our goal is to create something that is user-generated and brings together original audio recordings of poems of all ages and cultures. For the project we used the JavaScript framework React and Firebase as the database. Our beta website contains many key functionalities, such as recording spoken poems, searching for recorded poems, and listing recommended poems. It has an organized and clear visual appearance that should appeal to potential users. In addition, our team collected market data and user information to help define website features and methods to promote traffic to the site.

 

 

Expandi Front-end: Porting a complex multi-player board game to the Web

Presenters: Anthony Panisales, Benny Fung, Emmit Parubrub, Joshua Chong

Sponsor: Jon Rahoi

Faculty Advisors: Profs. Doug Halperin, CS490

 

Our project, Expandi, is an online, multiplayer adaptation of the German board game “Gaia Project.” The game involves complex strategies in the pursuit of gaining inter-galactic territory on a complex board. The physical Gaia Project board game has many challenges with its gameplay including a lengthy setup time, significant difficulty implementing randomness, and trouble with tracking game values. Implementing the game on-line (as Expandi) offers solutions to those problems. Using a React framework, our team has focused on addressing these challenges through computerizing otherwise manual functions while presenting a beautiful front end for this intriguing board game

 

Property Management: Building a platform to connect renters and managers
Team: Ken Ahrens, Sean Bowman, Ariana Jorgensen, Steven Worrall
Sponsor: Jose Alvarado
Faculty Advisors: Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

 

The Web application we built is a comprehensive platform through which renters and property managers alike can accomplish the myriad of tasks involved in property rental. For example, tenants can search through available properties, apply to rent properties, pay their rent, or submit maintenance requests. Meanwhile, property managers can post their available properties for rent, view tenant applications, handle maintenance and repairs, and keep track of financial activities. This Web application, built on a stack of technologies from PosgreSQL to React, aims to ease friction in the process of renting properties for both user groups by minimizing the number of websites or services needed to perform most tasks surrounding property renting. Market research and competitor analysis were done to suggest features to include in the app and to identify potential competitive advantages.

 

 

FANmire: Linking celebrities, fans and brands

Presenters: Arthur Alvarenga, Paolo Bondi, Tristan Perez, Talia Simanski

Sponsors: Jamel Anderson, Sean Dagony-Clark, Jason Tseng, Gabriel Brown,

Kacy Charles, Farid Hossain, Kim Moore, Courtney McGraw

Faculty Advisors: Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

 

FANmire is social media platform dedicated to creating a community where fans, influencers (celebrities) and brands can benefit from meaningful interactions with each other. The platform will give influencers the opportunity to optimize their social media presence and monetize exclusive content. Using the framework Vue, our team’s project was to design and develop an analytics dashboard to help influencers visualize their data and gauge the success of their pages. The team also helped with the re-design of the look-and-feel of the site’s mobile presentation. Additionally, we conducted market research and competitor analysis, and developed strategies to aid in the launch of the platform.

 

Sudoku School: Game-playing with instructional support

Presenters: Alexander Oh, Derrick Shiu, Evan Fancher, Phillip Nguyen, Stewart Chiodo

Sponsor: John Huang

Faculty Advisors: Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

 

Sudoku School is an application that allows users to play Sudoku and is original in its ability to help players learn different gameplay techniques. For our project, we developed the foundations of a Web-based application and did a rebranding of Sudoku School. The developers, using React, focused on the user interface; the business team focused on market research and competitive analyses of other Sudoku apps in the market to develop rebranding concepts and monetization strategy recommendations.

 

 

Machine Learning: Enhancing user engagement

Presenters: John Murray, Kelsea Flores, Nick Perez, Vera Lobkina 

Sponsors: Roger Ruttimann, Reactful 

Faculty Advisors: Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

 

Reactful is a web-optimization software designed to analyze and respond appropriately to on-line behaviors. The overall goal is to facilitate and encourage users to continue interacting rather than abandoning a site because of confusion or other reasons. Our project was to build a predictive Machine Learning (ML) system to identify appropriate user cues and determine how the Reactful system should respond to them (for instance, present a modal window or show a pop-up) to encourage continued engagement with the site. The model is being built on the Google Cloud Platform in TensorFlow around Reactful’s database of user behaviors and conversions (e.g., purchases). The model will be A/B tested in a production environment. It will be considered successful if it yields a higher level of conversions than the previous model.

 

 

Match My Template: Simplifying management of electronic signature documents

Presenters: Milan Chovatiya, Pragya Garg, Sarah Lopez,

Sreedeepraj Ratnasabhapathy, Sophie Wilton

Sponsors: Pooja Brown and Michael Darmousseh, DocuSign

Faculty Advisors:     Olga Karpenko, CS 690

Profs. Doug Halperin, CS490; Luigi Lucaccini, BUS 379

 

The team set out to improve the template-matching algorithm currently implemented by DocuSign, a leader in the electronic signature industry. The team developed independent document-matching algorithms (TF-IDF, Metadata, Latent Semantic Indexing, Locality Sensitive Hashing) and used the scores of these algorithms to train a neural network that produced the final document-matching score. This was delivered as a service architecture that allows for matching to run independent of the DocuSign system.

 

 

Secure Computing: Increasing security at the hardware level

Presenters: Brian Bushree, Marcus Chong, Noah King, Theresa Nguyen

Sponsor: Prof. Paul Lambert
Faculty Advisors: Profs. Doug Halperin, CS490

 

Security has typically been given greater emphasis at the software level. However, today with the increased number of software vulnerabilities and poorly implemented software, security must be implemented from the hardware level itself which serves as the foundation of most devices. In our project, we are using a secure element to secure cryptographic keys used to implement end-to-end encryption for a messaging platform. Some beneficiaries of this application are privacy proponents and journalists. For the messaging aspect of the system, we are using Apache Kafka as the message broker. Our system design is protocol independent and platform independent. It means that we can use any message broker or extend the design to support other features beyond messaging. Some examples of other applications are cryptocurrency hardware wallet, multi-factor authentication (MFA) device and IoT controller. This is because in all the above applications, the cryptographic keys are protected by the secure element. In our project, we chose to use Zymkey as the secure element which is designed for the Raspberry Pi which serves as a client in the system.