I'd like it if you come up with your own ideas! However, if you'd like to get some ideas, here are a bunch of suggestions. While most of these are a bit more free-form projects (mostly design studies) we also provide some paper re-implementations or re-evaluations.
Gaia is an ambitious ESA mission to chart a three-dimensional map of our Galaxy, with a first data release towards the end of 2016. Gaia will provide unprecedented positional and radial velocity measurements with the accuracies needed to produce a stereoscopic and kinematic census of about one billion stars in our Galaxy. The Astronomical community has never been faced with such a exciting but challenging task.
The goal of this project is to get a first contact with the newly released astronomical data base and perform a clustering analysis towards a search for new types of astronomical objects or groups of objects. This will be done on a sub-set of the full data, for which distances are known (about 2 million stars). We will first identify known groups of stars, extract parameters, and search for so far non identified, or previously ill-defined objects or structures.
Spreadsheets are arguably one of the most used computer tools in todays world -- not just in a financial and budget setting, but also for scientific as well as personal purposes. However, one of the most crucial consideration in understanding and predicting budgets is the uncertainty of the outcomes. Whether it is the financial forecast of a company, the budgeting of conferences and workshops, ... incorporating the variance of possible scenarios is cumbersome to impossible. In this project you will create FuzzySpreadSheets which will incorporate an elaborate uncertainty analysis in a common spreadsheet with a focus on usability, proper visual encoding, computational concerns, as well as cognitive aspects. Currently a spreadsheet cell holds just one numerical value, a FuzzySpreadSheet cell will hold a number, a set of numbers, an interval or more generally a probability distribution function. This requires some thoughts on how to specify the complex cell contents in an effective and user-friendly way.
Histograms are often used as the first method to gain a quick overview over the statistical distribution of a collection of values, such as the pixel intensities in an image.
Depending for example on the datatype of the underlying data (categorical, ordinal or continuous) and the number of data values that are available, several visualization parameters can be considered in constructing a histogram:
The perception of a histogram might vary quite a bit depending on the exact parameters chosen, and this might also influence the interpretation. On some of the above points, you should be able to find literature already.
The lecture on Algorithm and Data Structures includes an assignment in which each student has to implement a data structure for sorting. At the end of the course, it is possible to compare one's own results with everybody else. However, there is a large set of graphs that are being produced and one looses the overview very quickly. The task is to create an interactive program to create a better overview of all the data as well as good interaction techniques to quickly drill down into the relevant details a user would like to see. (download data, 365KB)
Latex is a very common document formatting language used for writing academic papers, grants, and reports. The default layout parameters of LaTeX result in very nicely formatted text but do not take into account final page length. Often one of the major limiting factors of these publications is page length. However, often papers are first written and refined without considering this limitation and then edited down to fit in the length requirement very close to the deadline. Because LaTeX's layout algorithm is somewhat unclear how it will adjust document length to changes in the source text and formatting paramters. Therefore, writers must resort to a long process of trial and error changing document content, adjusting formatting parameters, and recompiling the document in order to see if the document is within the page limits and still aesthetically acceptable. With some heuristics of document layout plus parameter space exploration and visualization of the final results it would be much clearer how the document will change this would allow multiple changes to be made at once and fewer recompilations. This process could be made much faster.
The idea is to support your fellow students in the Computer Graphics class. One of the most difficult things here is to debug your code. One way to do this is to try a lot of different parameters in an efficient way. E.g. lots of different material parameters, different light intensities, different camera placements and so on. Imagine you had such a tool which would create lots of images with these different setting and quickly lets you browse through your results. Wouldn't that be great? Well, you could create one such tool!
Long term aim of the project is to develop an instrument, able to monitor the latent transaction processes in group communication. Extending the concept of sentiment analysis, the project is located on the intersection of latent semantic analysis and psychoanalytic theory. A group is a defined set of actors coming together to realise a predefined group task.(work-group) Interactions within the group are more often than not sidetracked by latent sentiments and desires, such as anxiety, envy, greed, narcissism. The group thus is hijacked, transformed, paralysed. In corporate environments these tendencies are accountable for substantial financial losses. What we learn from psychoanalysis is the distinction of manifest and latent content: What ever we do or say, there is a part which lies beyond our control or perception, yet influence our behaviour in groups.
The challenge here is a reverse engineering: if group communication is understood as symphonic piece, which follows certain rhythms, harmonies, dependencies, how could the annotation of such a piece be conceived, the underlying principles made visible? Transcripts of group interactions consist of a vast amount of text, the processing of which poses a challenge and demands creative ways of exploring and visualising currents within the text data for evaluation.
The project is based on therapy transcripts collected in a weekly group therapy session and thus represents a realistic live communication scenario. The weekly sessions with constant actors allow for a long term study and comparison of micro and macro variations. From an analytic point of view, the diagnosis of current condition of each group member, the current state of affect, cognitive capacity, capacity for concern, aggression, capacity for insight, extraversion, withdrawnness are variables of interest.
One criterium in the work with the data is the signing of confidentiality form, and the maintenance of a professional attitude towards the sensible nature of the data.
Dataset: anonymised group therapy transcripts 30 transcripts of 90 minutes conversations, current seed: approximately 70 pages. Dynamics: seed weekly appended by new session transcript.
Keywords: qualitative text interpretation, sentiment analysis, latent semantic analysisOne of the difficulties with machine learning is to really understand how an algorithm/algorithm family works. The goal here would be to pick a particular algorithm/algorithm family and help the user to better understand it by visualizing their behaviour. One way (but a promising way) is to expose their parameters and create lots of different results of the algorithms by varying these parameters. The summary of the results gives an overview of what this "black box" is capable of. Pick your favourite algorithm/algorithm family (SVM, clustering, Deep Learning, Neural networks, etc.) and develop such a tool.
Of particular interest to us are
In the last years Deep Networks, a special kind of artificial neural network with many layers, have revolutionised many fields such as Natural Language Processing or Computer Vision.
For image classification the Deep Networks are able to distinguish 1000s of different classes, unfortunately it’s not always clear for which type of class (e.g. dogs) the network works better and for which it doesn’t. In classic machine learning there’s the concept of confusion matrices which are a way to organise classification and mis-classification results in a simple matrix. While standard visualizations of these matrices are still usable up to about 12 classes, they unfortunately won’t scale up to matrices of size 1000x1000 as encountered in modern Computer Vision datasets.
Your job is to create new visualisations that scale to very large confusion matrices and enable an computer vision expert to understand the classification accuracy of his current algorithm, i.e, a convolutional neural network.
The Shadows We Cast is a ten minute ballet performance with real-time heart beat visualization of six dancers. All dancers wear chest belts to measure their heart beats. During the whole performance their heart beats are logged and visualised as typical EEG behind them on stage. As the dancers enter and leave the stage according to the choreography to dance alone, form couples or dance in groups, their heart beats vary throughout the whole piece. Furthermore, only heart beats of dancers on stage are visualised. This time and space dependent heart beat data should be meaningfully visualised for data exploration beyond using a typical EEG.
The data: The exemplary dataset contains the logged heartbeats and each line represents one heartbeat. Every logged heartbeat (e.g. 2017-05-07 19:58:46.475: E) has a timestamp (e.g. 2017-05-07 19:58:46.475) and a capital letter that identifies the dancer (e.g. E).
A selection of tasks to analyse:
The CoConUT project (http://coconut.cosy.wien) features smartphone apps which collect sensor data (location, speed, noise, nearby Bluetooth devices, heart rate, etc.) during mobile field studies. Result is a time series which shows information about the context and possibly interesting events the field study participants encountered. These time series data should be visualized and enriched by meaningful analyses..
The data: The data set consists of sensor data which is collected each second during a field study on the participant’s smartphone. The app can be downloaded from the App Store and you can create your own data sets: https://play.google.com/store/apps/details?id=at.ac.univie.cosy.coconut
An exemplary data set can also be downloaded here: https://homepage.univie.ac.at/svenja.schroeder/2017-cocovis/cocovis.json
A selection of tasks:
There has been a deluge of open data by various government and governmental organization over the last few years. While this is admirable, what good is all this data doing if the common citizen is not being able to understand, explore, nor learn from this data. Hence, the goal is to develop a tool (ideally) web based that helps people to explore such data. One of the challenges will be to gear this tool toward a broad set of people, hence you cannot assume a great visual literacy (a problem the New York times has been struggling with and perhaps is providing some ideas for). Further, it is unrealistic to provide a universal tool where all types of data can be explored with and all questions can be answered with. Hence, it'll be important to narrow your focus on specific aspect of civic life. There are quite a number of open data sources that you can choose from:
For a number of years, the Vis, InfoVis, and VAST conferences have created a visualization contest. For each contest a problem scenario together with the relevant data sets have been provided to the research community and a price has been awarded to the best visualization. Some of the problems have been quite challenging. However, for the most part, these are great problems to work on.
While we are not expecting you to create winning entries to these visualization challenges, these are often well thought out problems that are fun and solvable. See whether any are of interest to you.