Beta's projects for final year students 2015—2016

General

Note that the Final Year Project is your project, rather than your supervisor's.
Students are discouraged from making a choice blindly without discussing with me first. Please arrange for a discussion by emailing to clyip at cs dot hku dot hk.
Students applying for the projects please email me a statement why you are interested in the project, preferably with a CV for reference and records before selection.
This page may be updated from time to time, before the project selection exercise ends. So, check back from time to time.
The "Deliverables" section list out those deliverables explicitly required in addition to the standard deliverables (reports, source code, etc.) required by a HKU CS project.
Self-proposed topics welcome. Interested areas: meteorological computing, mobile applications, multimedia, computer music processing, perception, audio processing, neural networks, and artificial intelligence.

Quick links to projects

Natural science survey support system
Musical instrument recognition
Typhoon track predictor
Swift 2-based application design and development
Big data on aviation weather
Financial data forecaster

Natural science survey support system

Project info

1—2 students, maximum 2 groups for 2nd-major (6 credits) | Development project | Updated 2015-08-10[1]

Description

Natural science projects often involves recording the time and location particular species (e.g., birds, trees) or specimens (rocks) are recorded, as well as observation notes such as their behaviour (foraging, resting), measurements (juvenile bird, 5m tall tree) or conditions (weathered rocks). While the traditional way of keeping records on paper or cards is time-honoured and reliable, information such as the exact time location of records are not easy to keep.

This project is about making a system that would help surveyors to keep records in a convenient and reliable way.

Requirements

Can do some web programming.
There is no restriction on the language of implementation, but to make the system practical for surveyors on the field, use of web languages such as HTML, CSS, and ECMAScript/JavaScript with some server support is preferred, as the system can be run on web browsers on mobile phones locally. Communication to a server for submission or archival of records can be done in the background in a best-effort basis, as network connectivity may not always be reliable on the field.

Deliverables

A system is delivered that supports natrual science surveys.
Here are some possible feature of the system. It is not a requirement that all should be implemented.
- Search from a species/specimen list for entry.
- Automatic record of date of entry.
- Automatic record of location of entry.
- Calculating the accumulative total from incremental counts.
- Automatic data backup.
- Allowing the user to add notes to an observation.
- Allowing the user attach photos to an observation.
- Export of data to CSV format.
- Search of records of a particular species.

Musical instrument recognition

Project info

1—2 students, maximum 1 group from CDA (6 or 12 credits), 1 group from CS (12 credits), 2 groups for 2nd-major (6 credits) | Research and development project | Updated 2015-08-10[1]

Description

Human beings are good at recognizing sounds of various types. On phone you can recognize it's the voice of your friend. When listening to music, you can distinguish that a piano, a violin and some drums are used.

The project is to design and implement algorithms for recognizing the instruments used in a piece of MP3/AAC music input by the user and showing the list of instruments used. Ideally, the pitches of individual notes of the instruments can be extracted as well.

Requirements

Have some knowledge in music.
Interest and/or knowledge in audio signal processing, which involves quite a lot of mathematics.
Be able to learn pattern recognition methods.

Deliverables

The algorithm and a software system implementing the algorithm.

Typhoon track predictor

Project info

1—2 students, maximum 2 groups from CDA (6 or 12 credits), 1 group from CS (12 credits), 2 groups from 2nd-major (6 credits) | Research and development project | Updated 2015-08-10[1]

Description

The Best Track Archive of Joint Typhoon Warning Center (JTWC) contains about 60 years of data of the tracks of tropical cyclones all over the world. The Severe Weather Information Centre and Meteoalarm contain current data for weather warning. These data are publicly available and can be analyzed for applications based on statistics.

The project is to use these data to answer some questions, such as

The amount of time it takes for a tropical cyclone to dissipate given its location.
Whether a tropical cyclone at North Pacific Ocean would
- affect Taiwan.
- affect Japan.
- enter South China Sea.
- affect Hainan penisula.
- affect Vietnam.
- become exotropical cyclone.
That is the distribution of maximum wind level at a city a tropical cyclone born at a specific location would generate.
The probability of different kinds of wind generated at Hong Kong because of a tropical cyclone at South China Sea.
...

The student is expected to come up with questions similar to these above that makes sense meteorologically, and answer them by analysing data.

Requirements

Interest in meteorology.
Interest in data analysis.

Deliverables

A system that integrates and visualizes, and animates existing data.
A system for answering queries such as those above, that can be trained using updated data every year.

Swift 2-based application design and development

Project info

1—2 students, maximum 2 groups from CS (12 credits), 2 groups from 2nd-major (6 credits) | Development project | Updated 2015-08-10[1]

Description

Now that the language Swift 2 is out and open-sourced, it is expected to become more popular. The project is to develop an iOS application by look into the details of the language and taking advantage of the language features.

It is open to the student to design and implement a Swift 2-based application or application suite to be run on watchOS, iOS or OS X.

Requirements

Interested in development using the Swift 2 language.
Have the necessary devices for development and testing.

Deliverables

The application and its source code.

Big data on aviation weather

Project info

1—2 students, maximum 4 groups from CS | Research and development project under HKU CS — HKO collaborative training agreement | Updated 2015-06-01[1]

Description

Convective weather systems such as thunderstorms entail risk of extreme turbulence, hail, wind shear, downbursts and more. According to statistics by International Air Transportation Association on weather-related aviation accidents from 2010 to 2014, the contribution of thunderstorms and wind conditions are 22% and 51% respectively. In Africa, thunderstorms’ contribution is a startling 88%.

Under significant convective weather, pilots and air traffic controllers have to make time-critical decisions to reschedule/vector aircraft on flight paths to maintain safety and ensure air traffic efficiency. Their decisions affect air traffic management parameters such as airport acceptance rate (number of arrivals an airport is capable of accepting an hour), flight delay and fuel consumption.

The goal of this project is to identify the relationship between convective weather and air traffic management parameters over Hong Kong using big data analytics, which enables discovery of hidden patterns and unknown correlations from huge volumes of data.

Weather data and aviation data will be given to groups taking the project for analysis using one of more of the following methods: classification methods, association rule mining methods, clustering methods, neural networks methods, etc. Each group will be using a different method for the same task.

Requirements

Have an idea how to interpret an weather radar image sequence in the Hong Kong Observatory (HKO) web site.
Have an idea what aviation weather services HKO offers.
Be able to write and/or learn some of the following languages, tools, and frameworks: Python, R, Java, Matlab/Octave, Weka, Hadoop, Spark, HBase, Cassandra, MapReduce.
Comfortable with the POSIX / Linux / Mac environment.
Not be daunted by system-level tasks like setting up network nodes fo computation.
Enter into an agreement for data access.

Deliverables

A software library for extracting features from meteorological data, such as radar images.
A scalable and easily portable implementation of generating the analytics using the chosen method.
The resulting relationship between weather data and air traffic management parameters.

Financial data forecaster

Project info

1—2 students, maximum 2 groups from CDA, 1 group from CS | Research and development project | Updated 2015-05-04[1]

Description

Given the historical data of a number of stock or index prices at different points in time, design an algorithm that would predict their values in the future.

Some factors the algorithm can take into the consideration include the day in month, weekday of day, time of day, various financial indicators, correlations between data from different time series.

Time series data of at least 15 stocks, futures and/or indices are expected to be analyzed.

Note that the student is expected to build their own collection of training and testing data.

Requirements

Interest and/or knowledge in financial data processing.
Interest and/or knowledge in data mining algorithms and tools.

Deliverables

A software system implementing the forecaster.
Sets of data for training, testing, and system parameter tuning.