Project Title: DEFINITION
EXTRACTION FROM CODE OF FEDERAL REGULATIONS
Summary: The
aim of the project is to extract definitions from the Code of Federal Regulations
using rule based approach and machine learning.
The project focuses on classifying paragraphs from Title 7 of the Code of Federal
Regulations. The algorithm for parsing, rule-based grammar and machine learning
is generic so as to support extraction of definitions from other Title, ideally
without modifications.
Course number: CS
5740, CS 4780
Faculty Advisor: Claire
Cardie
Topics
Artificial Intelligence
Engineering
Other
Presenters
Neha Kulkarni
Deepthi Rajagopalan
Mohammad AL
Asswad
|
Project Title: Speare
Summary: As
the publishing industry continues to shift to the digital space, news and media
sites have access to far more data about their readers. However, they have failed
to properly harness it. Speare is an intelligence platform for news and media websites
that takes advantage of this opportunity by using their semantic data to engage
and understand their audience.
- - - - -
More Information:
Speare creates intelligent user profiles of site visitors based on what they read,
and how they read it. We use Natural Language Processing to map a reader’s
site behavior and demographic information to their interests (concepts, places,
people, entities and topics). With knowledge of their interests and reading behavior,
we are able to provide two critical solutions: content recommendation, and actionable-insight
for the publisher.
First, we make sure that the right content from the publisher gets put in front
of the right reader. We do this by using their user profile, which is constantly
refined over time using Machine Learning, and our in-depth understanding of the
publisher’s content.
Furthermore, this same data fuels the Speare analytics dashboard - the first user-centric
and semantic analytics platform. With our data segmentation, a publisher can search
for something like “show me data on all repeat users that are interested in engineering
or automobiles and live in Ithaca NY.” This would pull up all past data on that
unique user segment. All of this information is then made visual, and actionable,
for the publisher.
Topics
Artificial Intelligence
Database and Information Systems
Theory, Math, and Modeling
Presenters
Rahul Shah
Alisha Sojar
|
Project Title: Predict
review quality on Yelp
Summary:
Course number: CS5999
Faculty Advisor: Thorsten
Joachims
Topics
Artificial Intelligence
Presenters
Khushboo Pandia
|
Project Title: Rhythm
Summary: Rhythm
is an Android application that helps users take a goal oriented approach towards
maintaining healthy patterns and inner circadian rhythm. It helps formulate their
sleep goals and work towards them, by providing active feedback and actionable suggestions.
Novelty of this application is the innovative method of collecting comprehensive
sleep data and providing relevant the user relevant feedback/mechanism for the user
to work towards a given health goal.Although other smartphone apps are available
that allow the user some degree of sleep tracking, most of them are obtrusive in
some way to the user.
Rhythm will assist individuals in tracking their rhythm and provide feedback, motivation,
and incentives to work toward and maintain a healthier rhythm. Now under development
is a Pebble smart watch app to help users better track their sleep and access Rhythm
in a more dynamic way without interruptions.
Course number: INFO
6130
Faculty Advisor: Tanzeem
Choudhury
Topics
Artificial Intelligence
Database and Information Systems
Biological Sciences
Presenters
Siddharth Bajaj
|
Project Title: Elastic
Machine Learning Compute Engine
Summary: To
enable users to analyze their data using standard Machine Learning algorithms out
of the box. Instead of a user configuring an Amazon EC2 instance with his application
code and running his algorithm, we will provide this solution through a simple and
configurable dashboard. The user has to simply select the algorithm he wants to
run through a dashboard, the data on which he wants to run the algorithm and the
parameters (if any). The user does not have to write any application code . Hence,
this is a cloud based solution, specifically a "Compute as a Service" infrastructure.
The application will be built to be highly scalable, elastic, reliable and fault-tolerant.
-----More Information-----
Our technology stack consists of a custom built Apache HTTPD 2.4 server with our
custom designed load balancer module. The application servers will be running Apache
Tomcat 7. Hadoop Distributed File System is being used for our persistent storage
needs
Features
------------
>A variety of machine learning algorithms to choose from via a Dashboard
>Ability to add custom datasources like AWS S3 or the user can upload his data
to our servers. We use Hadoop File system(HDFS) for storage purposes. HDFS will
be configured for High Availability.
>User can add/remove the number of instances for computation providing Elasticity
>User can generate reports and store them. User will be able to export his results
as CSV or JSON or even JPEG images of charts.
Course number: CS
5412
Faculty Advisor: Kenneth
Paul Birman
Topics
Database and Information Systems
Artificial Intelligence
Other
Presenters
Krishna Sasank Talasila
Preetam Shetty
Pritam Patil
Roshan Shetty
|
Project Title: Cloud-based
Smart Building Occupancy Tracking
Summary: To
significantly reduce in the energy usage of building lighting systems and improve
occupant comfort, we propose a reliable, fault–tolerant, consistent cloud-based
system for real-time tracking of occupant movements inside buildings and adjust
lighting systems dynamically. The clients periodically scan for all the Wi-Fi Access
Points it can see from a given location along with the relative signal strengths
and send that information to the backend server via TCP Socket.
The backend servers receive multiple messages from different clients, parse the
message to extract different fields, especially the Received Signal Strength Indicator
(RSSI). The server determines the occupant location by using the weighted centroid
algorithm. We use Isis2 to enable servers to process information captured by users
in the whole city at the same time and return a three dimensional location immediately
for every occupant.
We have a live demo to simulate multi-clients and lighting control system in the
city.
Course number: CS
5412
Faculty Advisor: Ken
Birman; David Albonesi
Topics
Database and Information Systems
Engineering
Presenters
Qianying Zhang
Shuo Xiong
Yang Yang
|
Project Title: teamwork.JS
Summary: Teamwork.JS
- Framework for distributed computing on web browsers
#WHAT IS Teamwork.JS?
Any web browser displaying a web page with teamwork.js becomes part of a bot-net.
Each node in this bot-net has two-way communication with the master and will run
any JS code distributed by the master.
Teamwork.js is a framework for distributed computation using the resources of website
visitors
#BENEFITS
As alternatives to native programs, web-browser based solutions have lower entry
barriers. Volunteers can join the system by just opening a URL; or be forced to
join by appending the JavaScript code to the html pages served (think ad-networks
and proxy services).
This means the feasibility of having millions of free compute nodes. Think of the
possibilities: precisely what motivates us.
#BACKGROUND
Modern web browsers are capable of handling extensive computations (V8 JS benchmarks)
as well as perform real two-way communication (WebSockets) and peer-to-peer communication
(WebRTC). This makes them capable of being used in such distributed settings to
solve parallel problems.
#BOOM
At BOOM we demonstrate the working of our system in a password cracking application.
#MORE
visit: http://teamworkjs.com
Course number: CS
5412
Faculty Advisor: Ken
Birman
Topics
Database and Information Systems
Presenters
Hammad Malik
|
Project Title: Adding
Infiniband Support to the Isis2 Library.
Summary: Infiniband
is a high performance server interconnect technology that is used as an alternative
to Ethernet. This project aims to enhance the Isis2 Cloud Computing Library to directly
work with Infiniband hardware, which should vastly improve performance and reliability
on these systems. To do so we create a wrapper for the libibverbs implementation
of the Infiniband Verbs API that can be called by the main Isis2 Library.
Faculty Advisor: Ken
Birman
Topics
Database and Information Systems
Presenters
Jonathan Behrens
|
Project Title: Accessing
Isis2 functionalities from a CLI and C++ client
Summary: The
objective of this work is to support and extend the functionalities of the stand
alone server which allows a C++/CLI client to access OOB Isis2 system calls, Isis2
locking layer and the Isis2 DHT. This server is basically a WCF Restful service
whose interface is exposed through a URL.
Course number: CS5412
Faculty Advisor: Prof.
Ken Birman
Topics
Database and Information Systems
Presenters
KAVERI CHAUDHRY
|
Project Title: Airport
Runway Scheduler
Summary: Most
airplane flight delays are caused not because a plane is not ready to depart by
its scheduled time, but because the airport's runways are unable to accommodate
it. Most airports only schedule runways according to their own local information,
so we propose a distributed global system that optimizes runway assignments.
Course number: CS
5412
Topics
Database and Information Systems
Artificial Intelligence
Presenters
Chris Heelan
Alex Maass
Gary He
|
Project Title: Smart
Cars
Summary: We
want to build a software for smart cars to automatically drive themselves to a destination
on a predefined route without colliding with other cars. The significance of this
project is that we will have a scalable system that coordinates the actions of multiple
cars to make them drive safely to their destination without collision with each
other. Whenever there is a potential collision, the cars will intelligently route
themselves so that they will make progress with no deadlocks.
We will give a live demo during which participants can select any desired path for
the simulated car to move along and the car moves along that path without colliding
with cars in its vicinity.
The project is a starting step towards promoting the idea of building smart cars
that can drive with no human intervention
Course number: CS
5412
Faculty Advisor: Ken
Birman
Topics
Database and Information Systems
Presenters
Jisha Kambo
Jessie Lin
|
Project Title: Pulso
Summary: Pulso
is a hand-mounted mobility device that allows visually impaired individuals to navigate
their surroundings and perform day-to-day tasks by “feeling” their way around. Pulso
uses both ultrasonic and infrared sensors to sense nearby objects and translates
this information to the user through intuitive vibrational pulses.
Pulso is used by sweeping the device side-to-side like a white cane, checking for
obstacles in the user’s path. Pulso provides extended range for improved obstacle
detection and allows the user to detect various common hazards such as approaching
obstacles and changes in elevation.
When used on nearby objects, Pulso uses an infrared-based ranging technique that
allows the user to accurately and efficiently detect edges and distances of objects.
This allows the user to generate a mental map of various objects, such as items
on a table, just by sweeping his or her hand across them. The user can also “feel”
his or her surroundings (especially in crowded and cramped places like public transportation)
in a non-intrusive and safe manner.
Course number: ECE
4760
Faculty Advisor: Bruce
Land
Topics
Engineering
Presenters
Eileen Liu
Shane Soh
|
Project Title: SARA
Summary: Patent
approved SARA (Search And Rescue Assistant) is the solution to emergency response.
Once activated, a car with SARA will send a voice message to 911 with the necessary
information immediately. The driver does not need to panic knowing that emergency
response will arrive soon. Unlike existing devices, SARA will convert the data and
send information in a voice message, which means that receivers would not need additional
devices to convert machine language to human readable texts for interpretation.
This therefore decrease the waiting time for drivers in accidents significantly
and can be easily commercialized as no country would need to allocate additional
funds to add devices to emergency response stations.
Course number: INFO
4400
Topics
Engineering
Robotics & Hardware
Human-Computer Interaction
Presenters
Jaeil Lee
Bok Young Kim
Tingting Wu
Theodore Lee
|
Project Title: Moxie
Summary: To
provide users with the incentive and feedback to maintain healthy lifestyle changes,
we envision a personal health application that monitors several dimensions of biometric
data, as well as additional user input to gain a holistic picture of a user’s lifestyle.
Our core product is a hardware-enabled app that uses data collected daily from our
smart scale to track and provide analytics for a user’s health. In addition, we
have developed a Wi-Fi enabled smart scale that quantifies a user’s health by noninvasively
recording a wide set of biometrics such as heart rate, blood pressure, and cholesterol
in several seconds while measuring weight and syncing to the cloud. We pair the
scale with a mobile app which logs and displays summaries of the measured data,
allowing users to judge the effectiveness and quality of their own health and make
changes accordingly. We increase the rate of data collection from once or twice
per year (average adult check-up rate under a primary care physician) to once per
day. Our goal is to have users measure their health the same way they measure their
weight in the morning to reduce the risk of preventable diseases. This results in
the following effects:
1. Goals and Milestones: Users can remain motivated to push harder on health programs
if the goal is always in sight.
2. Self-quantification: The device itself will allow users to quantify and qualify
the effects of different lifestyle choices, and track their own health outside the
doctor’s office.
3. Learn from others’ mistakes: Users can easily discover health treatments and
products that have been verified to work for their particular body type based on
confirmed success from other users.
By using our device, customers will be able to create custom solutions for their
unique problems and meet their personal health goals. With data on a large number
of users (anonymized data used with permission), we can infer what small changes
in user activity are effective in promoting sustained change. Our application involves
both data collection and analysis, and provides long term health intelligence for
both consumer and medical industry interests. We use the collected data to build
a platform through which users can quantify different strategies and predict how
different plans can work for them, based on data from similar users. This creates
a recommendation system that allows users to pick the most effective dieting and
health routines based on measured body type and background.
Course number: CS5412
Faculty Advisor: Ken
Birman
Topics
Engineering
Database and Information Systems
Biological Sciences
Presenters
Alex Ngai
Randy Song
|
Project Title: Gesture
Based Security Lock
Summary: The
aim of our project is to design a security system which can be unlocked by means
of a stored gesture pattern. The idea is to create a box like assembly, in which
the user places his hand, makes a defined gesture and unlocks the system. Basically,
there is a mechanism that allows the user to save a gesture pattern. Once that is
done, the system goes in lock state. When the user enters his hand in the box, he
tries to recreate the same pattern. If he is able to do so, the system unlocks.
If unable to, the system remains locked.
The idea is inspired by the popular mobile phone unlock feature where basically
you draw a pattern on the screen and the phone unlocks. We wanted to create a similar
system which could be used in any security application, as simple as opening the
door of the house based on the gesture. The attractive feature of the project is
that the user makes the pattern in the air and not on any surface. Also, we have
given the user the flexibility of changing the pattern whenever he wishes to do
so.
The whole idea is to provide a fully functional product to the user who can interface
the system to any electronically controllable device. The user can himself create
patterns, store them and open the device only by himself, giving him the freedom
to lock and unlock the device without any kind of key required.
Course number: ECE
4760
Faculty Advisor: Bruce
Land
Topics
Engineering
Human-Computer Interaction
Robotics & Hardware
Presenters
Ankur Thakkar
Darshan Shah
Saisrinivasan Mohankumar
|
Project Title: SECURITY
SYSTEM FOR DYNAMIC DATA SHARING OVER SCALED INFORMATION SYSTEMS
Summary:
Proposal:
Building a system that will provide a security solution for GridCloud, based on
a multi-domain security model. Whereas organizations would all use the GridCloud
platform each having multiple clients’ connections, but would have individual ownership
and control over their individual data. The System would allow explicit portals
to be opened between organizations, if both authorize them. Beside to a visualization
tool for understanding, who is authorized to do what, a representation of the security
rules and policies, keys to implement these rules will be developed.
Goals:
To provide security management across multiple domains, for a shared/hybrid cloud
infrastructure containing sensitive, critical data with following security principles:
1. Access Control
2. Protection against Data leak
3. Anomaly Detection
4. Load Balancing
5. Scalability
6. Crash Recovery
Analytical Steps:
1. Define Cloud Infrastructure
1a. Build System of Application Servers and Databases
I. Divide them into two/multiple domains, emulating Company A and B
1b. Build Clients:
2. Defining access control points:
2a. Building policy enforcement point (PEP)
2b. Building policy control point (PCP)
3. Defining access control mechanism:
3a. In PCP, create function to define and push policies to PEP.
3b. Defining communication protocols between PEP and PCP.
3c. In PEP, create function to receive and implement the policies from PCP.
Milestones:
• Phase 1:
I. Setup infrastructure
Creating application servers.
Creating databases
Creating clients emulating employee and subcontractors
Making the appropriate connections.
II. Create PEP and PCP.
III. Implement access control mechanism as defined in previous section.
IV. We must be able to connect one Client to the Cloud, with application of policies
and data encryption.
• Phase 2:
I. Create a GUI to PCP
II. Implement date leak prevention feature.
III. Provide restricted access control and short term access features to some specific
clients.
• Phase 3:
I. Provide mechanisms to identify and prevent Anomaly Detection.
II. Develop complete GUI for both PCP and PEP
III. Create a visualization tool for the applied policies.
• Phase 4:
I. Implement Load Balancing and Scaling mechanism for PCP and PEP
II. Implement Crash Recovery
Evaluation plan:
• Demo:
Setup the system
Simulate the clients: connect, authenticate, get encrypted data
(Verify accessing to data according to the policies which have been applied in CP
for specific client).
Delete/Add new policy, apply.
(Verify accessing to data according to the policies which have been applied in CP
for specific client).
The policies implemented at PCP should be propagated to all PEP
Create an anomaly and check if it is detected by the system, and protective action
is put in place.
Visualization: representation of the security rules and policies.
Run test cases:
Test1: Test for access control, all form of clients must be able to connect and
disconnect, with the application of the necessary policies.
Test2: Test for Protection against Data Leakage, test by accessing unauthorized
Servers/Databases, access should not be allowed.
Test3: Test by Simulating a DDoS attack, the system must detect that a attack
is taking place and put the appropriate protection in place either by shutting malicious
client down or some other mechanism [TBD].
• Poster:
Architecture of our system
Technologies used
Experiment results:
Access Control
Protection against Data Leakage
Encryption/Decryption
Anomaly Detection
Ganesh Gs563@cornell.edu
Sarah Saa254@cornell.edu
Sneha Sh824@cornell.edu
Course number: CS5412
Faculty Advisor: Ken
P Birman
Topics
Engineering
Other
Presenters
GANESH SEERPI
SATISH CHANDRAGUPTA
SARAH ALABDULLATIF
SNEHA PRASAD HEBBUR
NARASIMHA PRASAD
Rony Krell
|
Project Title: xRemote
Summary: The
xRemote is a small and inexpensive multi-device remote controller which can be controlled
by cell phone
application through Wi-Fi connection. It is designed for all types and brands of
infrared-controlled appliances
through reserved coding and self-learning. It allows user to control household appliances
anywhere internet
access is valid. The xRemote will introduce a new kind of lifestyle differing from
traditional remote control. It
will become an important milestone to enter the next generation Internet-of-Things
smart home technology.
Faculty Advisor: Bruce
Land
Topics
Engineering
Other
Presenters
Georald Xu
Yefei Shen
|
Project Title: Identifying
and Extracting Statutory Definitions from U..S. Code
Summary: Legal
statutes, rules and regulations are difficult to understand because terms mentioned
in legal corpora may have meanings that are very different from their general meanings
and senses. Unfortunately, it is very difficult to find the right definition for
a given term because: (1) definitions usually exist in pages or sections that are
different from pages or sections where these terms are used; (2) these definitions
are scattered across the corpus and are not organized in glossaries or appendices;
(3) there are many definitions within legal text corpora (Title 7 of United States
Code (U.S. Code) has over 1500 definitions) and (4) the same term may have different
definitions in different context or scope. To support better understanding of legal
text, it is very important to find these definitions and map them to where they
are used within the corpora. This project focuses on automatically extracting definitions,
their defined terms and scope (where the definition applies) from the U.S. Code
by utilizing a bundle of rule-based and statistical natural language processing
techniques. Every paragraph in the U.S.Code is treated as a candidate definition
and then a classifier that is trained on a sufficient number of correct/incorrect
definitions is used to classify paragraphs into definitions and none definitions.
Then we apply a number of well-defined rules to detect defined terms and scope of
definitions. Finally, we use named entity recognition and detection techniques to
find the defined terms in their context within their scope and map them to their
definitions.
Course number: CS5999
Faculty Advisor: Mohammad
AL Asswad
Topics
Engineering
Artificial Intelligence
Other
Presenters
Mohammad AL
Asswad
Arpana Hosabettu
Harsh Shah
Remya Balasubramanian
Yashaswini Shekarappa
|
Project Title: B33P
Summary: B33P
is a puzzle platformer game for the PC with unique, programming-like controls. Choose
a sequence of adjacent tiles from an action board to guide B33P on his adventure,
anticipating the movement of enemies and the environment.
Course number: CS
3152
Faculty Advisor: Walker
White
Topics
Gaming and Entertainment
Presenters
Kristin Murray
Athena Cole
Bella You
Heming Ge
Emily Lutz
|
Project Title: microPig
Summary: MicroPig
is a physics platformer iPhone game that features a unique game mechanic. Travel
through the levels by grappling on to berries and avoiding obstacles to save all
the fairies as the whimsical microPig.
Course number: CS4152
Faculty Advisor: Walker
White
Topics
Gaming and Entertainment
Presenters
Ruijun Wang
|
Project Title: Casino
Heist
Summary: Your
name is JACK STEEL. You received an invitation to Il Casino Bianco, and upon arriving
at the penthouse, you decided to steal a fortune. Your goal is to escape down to
the bottom of the casino with as much money as possible.
Course number: INFO
3152
Faculty Advisor: Walker
White
Topics
Gaming and Entertainment
Presenters
Zach Porges
Laurence Rosenzweig
Ryan Hall
Tre' Calhoun
Nicholas Tyson
Keith Newman
|
Project Title: Apsis
Summary: Apsis
is a playful, uplifting experience about flying built for Android tablets and PCs.
Players shepherd a flock of birds through a hand-painted world and experience an
emotionally rich story narrated entirely through gameplay.
Course number: CS
4152
Faculty Advisor: Walker
White
Topics
Gaming and Entertainment
Media, Art, and Graphics
Human-Computer Interaction
Presenters
Matt Blair
John Oliver
|
Project Title: Pirates
of the Stratosphere
Summary: Pirates
of the Stratosphere is a 2D sidescrolling shooter game in which players take to
the skies for a highflying, pirate adventure filled with ship combat, commandeering,
and plundering. Fight through intense bullet-hell ship-to-ship combat, commandeer
variety of enemies to steal their ships, and explore beautifully rugged pirate worlds.
Course number: CS
3152
Faculty Advisor: Walker
White
Topics
Gaming and Entertainment
Media, Art, and Graphics
Human-Computer Interaction
Presenters
Eileen Liu
Andrew Noyes
Peter Zieske
Ian Panzica
Matthew Hayes
|
Project Title: Sentiment
Design
Summary: The
MoodLight system explores the potential of an interactive system to facilitate reflective
mental health practices. Electro-dermal activity (EDA) sensors read biometric data
about an individual’s current arousal level; this information is fed into an interactive
lighting system; and fluctuations in arousal level are interpreted as changing colored
light.
One of our key interests in running a recent design probe was to explore the possibility
of supporting healthy social interactions about stress management through the deployment
of technology rather than in spite of it. Our study included a number of opportunities
to observe social interactions about, through and as a result of using MoodLight.
Faculty Advisor: Geri
Gay
Topics
Human-Computer Interaction
Media, Art, and Graphics
Social Media
Presenters
Nicole Calace
Michael Elfenbein
Jacqueline Chien
Adam Shih
Jonathan Lee
|
Project Title: Machine
Translation in Computer-Mediated Communications
Summary: The
concept of machine translation is nothing new, and IRC (Internet Relay Chat) clients
have been around almost since the advent of the Internet. But what happens when
you take the two and combine them? Can we use today’s imperfect translation tools
and combine them with a chat interface to create a better platform for collaboration?
This is the objective of the Machine Translation in CMC project, a study on the
patterns of computer mediated communications between native and non-native English
speakers. The interface itself consists of two parts: a chat window built with the
traditional server-client model, and an interactive map that supports basic drawing
functions. Together, these tools support a series of experimental conditions designed
to explain if and how machine translation can have a positive impact on multilingual
CMC communications.
Faculty Advisor: Susan
R. Fussell
Topics
Human-Computer Interaction
Database and Information Systems
Presenters
David Hau
|
Project Title: Trickle
Summary: Many
people in college start to drink alcohol. As college students quickly learn the
immediate consequences of drinking too much, it is often much more difficult for
them to be cognizant of the more subtle effects of drinking alcohol. Likewise, most
people are not aware of trends in their lives that can lead them to drink excessively.
Our proposed survey-based application will help users learn about the potential
effects alcohol can have in their lives. Our application could also potentially
expose triggers for drinking excessively to users who might have a drinking problem.
By recognizing the triggers, users will be able to more easily avoid them, which
could help them control their drinking problems. We have our initial Android prototype
from last year ready at: https://play.google.com/store/apps/details?id=cornell.eickleapp.
We have made strides and improvements on the system, but will not be ready until
much later.
Course number: INFO
5900
Faculty Advisor: Tanzeem
Choudhury
Topics
Human-Computer Interaction
Database and Information Systems
Artificial Intelligence
Presenters
Alex Pan
Tingting Wu
Stacey Wrazien
|
Project Title: SAINT:
A Scalable Sensing and Inference Toolkit for Mobile Devices
Summary: At
the end of 2013, it was estimated that 1.43 billion people in the world use smartphones.
All these smartphones come pre-equipped with sensors that can inform fine-grained
behavior of the phone’s user. For instance, audio information provided from the
mic, in real time, can be used to monitor the effects of ambient noise on a sleeping
person. After obtaining this data and performing careful trend analysis, we will
be able to know the exact sound wave frequencies or types of external sounds needed
for a person to sleep optimally. This is just one example out of the plethora of
possibilities of the benefits of mining and processing mobile sensor data. Open-source
software enabling large-scale phone data sharing can have large implications for
health care services and can further grasp societal scale behavior understanding.
The goal of this project is to develop a robust and scalable open source Android-based
sensing and inference toolkit to allow third-party client applications to gather
and process smartphone sensor data, ranging from activity recognition to voice recognition
to speech recognition. This interface, called SAINT (Sensing and Inference Toolkit),
is an Android application that runs as a background process in a mobile phone. SAINT’s
easy-to-use client-server architecture enables multiple client apps to register
with the SAINT server and obtain mobile sensor detector data in real time. How does
the client-server model work?
All sensor/service (e.g. accelerometer) and activity inference produced in real
time gets pushed into an intermediary bus. This bus will then provide each registered
client app with the data corresponding to the specific detectors it has registered.
For example, if a client app registers with the activity recognition detector service
and another client app registers with the speech detection service, then the bus
will provide the activity recognition service data and the speech detection service
data to the corresponding clients in real time. Thus, the bus acts as a “server”
while the listening apps are the “clients.”
Client apps can connect to the server for sensing and inference data from phone
detectors, or any entities that produce a real time stream of data. SAINT comes
with physical activity and voice recognition capabilities pre-packaged on the server,
and the expandable architecture of SAINT server allows developers to easily add
new inference capabilities.
Faculty Advisor: Tanzeem
Choudhury
Topics
Human-Computer Interaction
Database and Information Systems
Other
Presenters
Shankar Athinarayanan
|
Project Title: Tailor-made
Games
Summary: Personal
Informatics has emerged in recent years to describe tools designed to help people
collect and reflect on personally relevant information. These tools go hand in hand
with the drive to change behaviour or improve oneself, most often referred to as
the Quantified Self.
This project is setting out to establish a platform for using personal information
in computer gaming in order to understand what the range of uses this approach might
have.
Most games use sparse amounts of personal information as part of play. Using information
in its broadest sense, the player interacting with the game may be considered a
basic layer of information. In some cases, for example video poker and many other
gambling games, player “information” has no real impact on the game outcome.
The addition of personal information into gaming is an opportunity to add more subtlety
to computer games (custom-made). Such games can might provide new ways to gain self-learning.
Some suggested uses might include:
1. Games to change behaviour (based on personal info)
2. Games to facilitate social connection (based on sharing personal information
– disclosure)
3. Games for personal insights / reflection (based on personal information)
4. Games to learn about yourself (based on how you play as opposed to personal information)
Over and above the immediate goals of this project, the resulting software platform
will make it straightforward to create personally meaningful, tailor-made computer
games to guide and lead personal development and behaviour change. Building on this
platform, future games could use automatically capture mood and stress levelPersonal
Informatics has emerged in recent years to describe tools designed to help people
collect and reflect on personally relevant information. These tools go hand in hand
with the drive to change behaviour or improve oneself, most often referred to as
the Quantified Self.
“Just remember” is one of the games in our series. A set of photos is displayed
on screen and the player has to match similar photos. By exploiting the fact that
people often love to sit down and look back at their own photos, we explore: whether,
compared to a random picture set, a set of personal pictures increases the players
concentration span.
Faculty Advisor: Mark
Matthews
Topics
Human-Computer Interaction
Gaming and Entertainment
Social Sciences & Humanities
Presenters
Shuo Xiong
Hammad Malik
|
Project Title: Video
to Textual Summary
Summary: This
project was done for an external client UserTesting.com, a California based startup
as part of the masters project. The goal of the project was to develop a proof of
concept for extracting highlight points in a video. The video records the experience
of the user thinking out loud as (s)he browses through a website. Speech to text
processing and sentiment analysis was used to find regions of frustation/distress
in the browsing experience of the user.
Faculty Advisor: Gilly
Leshed
Topics
Other
Presenters
FNU Tushar
|
Project Title: Bus
tracking application
Summary: Our
project is a Cloud Based mobile bus tracking application. The mobile app, allows
users to select a bus number and based on the users location, it will tell the user
where nearest stops are, the time the next bus will arrive, the location of the
bus, and predicted time it will take the user to walk to the stop. The application
runs on android and uses google maps to display the information. The application
is simulated on a "local cloud" that we have developed. The "local cloud" is multiple
virtual machines connected together on our local machine. This simulates a cloud
environment and deals with multiple clients by connecting them to different servers
that are running on different virtual machines. Our cloud is designed specifically
for fault tolerance.
Course number: CS
5412
Faculty Advisor: Ken
Birman
Topics
Other
Presenters
Alex Kittelberger
Christopher Jonathan
|
Project Title: Cornell
Student
Summary: Cornell
Student is an Android application developed for students in an effort to bring most
of the services that they use into a single mobile application. This app allows
them to view the timetable for their enrolled courses, their office hours and such
other details. It also provides location-aware TCAT bus information and also quick
access to frequent-routes which would be customized and specific to each individual
user. It further offers broadcast services by giving them an option to communicate
among all the users of the app. There are also quick links to more helpful information
about the university.
It’s planned to have a three layered architecture consisting of the front end built
for Android, the middle tier application server built using Apache Tomcat servlets
and the back-end MySQL database. The latter two layers implement state of the art
cloud technologies. Application servers and databases are made fault tolerant by
replicating them. This also serves the dual purpose of offering high availability
and load balancing when there are high number of client requests. But it further
brings up the challenge of maintaining consistency among the servers and the databases.
Such complexities are handled to make our system robust and reliable.
Course number: CS
5412
Faculty Advisor: Ken
Birman
Topics
Other
Database and Information Systems
Engineering
Presenters
Ranjit Murali
Manjunath Patil
Nikash Narendra
Kumar
|
Project Title: Implementing
Generics for Jif
Summary: Jif
is a security-typed programming language that extends Java with support for information
flow control and access control, enforced at both compile time and run time. Our
goal is to augment Jif with the language features required to support generic programming
design patterns found in many of today's languages (such as Java). Modifications
to the Jif compiler are made using the Polyglot extensible compiler framework.
Course number: CS
4999
Faculty Advisor: Andrew
Myers
Topics
Programming Languages
Database and Information Systems
Presenters
Matthew Loring
|
Project Title: Cornell
University Autonomous Underwater Vehicle (CUAUV) Project Team
Summary: CUAUV
is an undergraduate project team that builds an autonomous submarine each year with
the primary goal of competition at the annual AUVSI / ONR RoboSub Competition in
San Diego. The competition takes the form of an obstacle course which requires vehicles
to complete a number of visual, acoustic, and manipulation tasks all without user
input. For instance, submarines must be able to navigate through the course by following
a series of path markers, ram a colored buoy, shoot a small “torpedo” through a
target, drop a marker on a target, manipulate pegs on a pegboard, localize an acoustic
pinger, and grab and retrieve an object. In order to complete this course autonomously,
our software is divided into three primary areas: mission, which focuses on high
level logic in order to navigate through the course and complete tasks; sensors
& controls, which requires pose estimation and feedback control in order to
move the vehicle as commanded by our high level mission subsystem; and vision, which
involves processing information from our cameras in order to classify and localize
elements in order to complete the required tasks.
Faculty Advisor: Alan
Zehnder
Topics
Robotics & Hardware
Engineering
Presenters
Markus Burkardt
|
Project Title: CUAir
Project Team
Summary: CUAir,
Cornell University Unmanned Air Systems, is an interdisciplinary project team working
to design, build, and test an autonomous unmanned aircraft system capable of autonomous
takeoff and landing, waypoint navigation, and reconnaissance. Some of the team’s
research topics include airframe design and manufacture, propulsion systems, wireless
communication, image processing, target recognition, and autopilot control systems.
Faculty Advisor: C.
Thomas Avedisian
Topics
Robotics & Hardware
Engineering
Presenters
Joel Heck
Derek Faust
Andrew Knauss
Cory Pomerantz
|
Project Title: Checkers
Playing Delta Robot
Summary: This
checkers playing robotic arm built in a delta configuration was made in the Spring
2013 rapid prototyping class. It uses a webcam to identify the current state of
the board, an Arduino microcontroller to control the servos, and an electromagnet
to pick up and move individual checkers. A human interacts with the system via a
large red button that indicates to the computer that the user has made a move, signaling
it to complete a move of its own. It is constructed with both hand and laser cut
pieces of wood, and 3D printed specialized components designed for the system in
the rapid prototyping class.
Course number: INFO
4320
Faculty Advisor: Francois
Guimbretiere
Topics
Robotics & Hardware
Engineering
Gaming and Entertainment
Presenters
Ronnie Bunshaft
|
Project Title: Piel
the Interactive Desk Lamp
Summary: Piel
is an animatronic desk lamp installation that tracks a user's hand movement beneath
it in order to provide a unique interactive experience. The lamp moves about in
3-Dimensional space using a series of motor-operated control strings, somewhat like
a marionette. As a camera tracks hand movement beneath the device, the lamp orients
itself to point a focused light source towards the hand.
Course number: INFO
4320
Faculty Advisor: Francois
Guimbretiere
Topics
Robotics & Hardware
Human-Computer Interaction
Presenters
Benjamin Itzkowitz
Ross Hanson
|
GoToo_Official_Logo.pdf
Project Title: GoToo
Summary: GoToo
is a mobile application that serves to easily help users find and create events
in their area. The application has a constantly-updating 'Events Feed' that is customized
to a user's preferences, along with pages for 'My Events', 'Create Event', 'Explore
Events', and 'Settings'. There are several unique features for GoToo. Users will
only see events occurring in the next 48 hours for spontaneous users searching for
events to attend. Our system also integrates with Facebook so users do not have
to create events twice.
The system is currently in its pre-Alpha phase, as we are about to complete the
native iPhone version of the application. Our goal is to test and modify our application
in April based on user feedback. This summer, we will build the native Android version
of the application, which is useful for including Android users as well. We plan
on releasing the application at ClubFest in September 2014, where we can start building
our user base through on-campus clubs/organizations.
Course number: INFO
4900
Faculty Advisor: Dan
Cosley
Topics
Social Media
Human-Computer Interaction
Presenters
Laurence Rosenzweig
Adam Groner
|
Project Title: Achieve
Summary: Achieve
is a mobile web application designed to help groups of friends work together to
achieve their goals.
Faculty Advisor: Dan
Cosley
Topics
Social Media
Presenters
Zach Porges
Yundi (Lily) Gao
Yulan (Lannie) Miao
Brian Lin
Ben Shulman
Evan Long
|
Project Title: daapr
Summary: daapr
is a social media platform specifically dedicated to the sharing and discovery of
online content. daapr provides users with a feed that aggregates all the content
(i.e. links) shared by the people they are following on the site. At the same time,
daapr allows users to share comfortably to an audience of followers who have expressed
an explicit interest in what they are reading and watching.
The site’s highly-visual feed was specifically designed to make it easy for users
to quickly scan posted material and determine its relevance to them. Users can also
easily resurface all the interesting content they’ve posted in the past through
their profile page.
We recognize that social media sites are only valuable once they reach a critical
mass, so we are developing a browser extension that allows daapr users to share
easily both inside and outside of the platform. Specifically, we will make it possible
for our users to email anyone they know a beautifully-formatted article link and
preview from any web page. This will allow us to expose potential users to our product
and make daapr a value-add for anyone who signs up.
Here’s a quick video of what we’ve made so far: https://www.youtube.com/watch?v=kTsl8N1nzao
Course number: INFO
4900
Faculty Advisor: Dan
Cosley
Topics
Social Media
Human-Computer Interaction
Presenters
Alex Meyers
Aaron Schifrin
Connor Strynkowski
Brendan Miller
Austin Gage
|
Project Title: Dual
Screen Processing: Redefining The Practice of Watching Television
Summary: My
thesis investigates the effects of dual screen processing on the modern experience
of watching television. To do this, 180 participants watched an hour of television
of their choice, and then answered a series of questions indicating age, gender,
the title of the program they watched, and the network on which the show was aired.
They also identified their media outlet and the location in which they watched the
hour of television. Finally, participants rated their current moods on a scale of
one to five (unhappy to happy) after watching the hour of television, and they indicated
whether or not they personally thought that engaging in dual screen media consumption
enhanced their television watching experiences. The second portion of the study
used a reflective diary method to record messages sent and received by participants
that had to do with the content of the television they viewed. Next, I coded the
data in the messages to identify the types of messages, as well as what viewers'
motives were in using other media while they watched television. My thesis and research
seek to answer the question of whether or not the use and integration of several
types of media enhance the experience of watching television.
Faculty Advisor: Jeff
Hancock
Topics
Social Media
Social Sciences & Humanities
Presenters
Jaimee Pavia
|
Project Title: Papiya
Summary: Many
applications today have the need for a location aware, consistent, and real-time
data store in the cloud. For this project, we will build an API that allows such
applications to add objects to the data store and to retrieve previously added objects
in real time. The application can receive dynamic updates based on changes in the
data store. Possible applications of our project are a Humans vs. Zombies utility
and social media or event-planning apps.
Course number: CS
5412
Faculty Advisor: Ken
Birman
Topics
Social Media
Gaming and Entertainment
Presenters
Simon Li
Shela Wang
Ivan Kang
|
Project Title: Designing
Technology for a Social Impact
Summary: We
are developing speculative designs for technology based on ethnographic and historical
research from Jamaica, Newfoundland, and Iceland. Using a historical understanding
of technology and focusing on the global margins who have seized on technological
progress, our project seeks to critically analyze the idea of sociotechnical development
and conceptualize this through design.
One of our goals is to create a proposal for the new handbook of Digital STS, a
volume published to advance the understanding of digital objects, phenomena, processes,
and methods in Science and Technology Studies. We hope to gain new insight on our
headway at BOOM, where we will be asking for feedback and ideas.
Faculty Advisor: Phoebe
Sengers
Topics
Social Sciences & Humanities
Database and Information Systems
Other
Presenters
Luna Zhang
Robert Utsey
Erik Bonadonna
Benjamin Potts
Richmond Wong
Holly Domke
|