Registration

First time users will have to register, selecting MTOP2016 as research unit in the process. If you already have an account with SMIR, navigate to Group Administration and select "Join another research unit" to join MTOP2016. In both cases, you will have to wait until your registration is confirmed by an administrator

Which methods are called for?

• The data set consists for 27 subjects belonging to 3 different categories (healthy, patient category I or patient category II) each consisting of 9 subjects.
• Goal of this challenge is to distinguish the subjects in one category from the other two in an unsupervised way.
• The proposed method should be fully automated and ideally computationally efficient
• Participants are required to provide one of three category label for each subjects
• Additional we ask for a probabilistic value that describes the confidence of a subject's association with the suggested category
• There is no restriction to novel and unpublished methods, as participants are likewise encouraged to submit innovative approaches as well as their existing methods.

Evaluation

Participant will have to assign a cluster label (1/2/3) to each subject. Methods will be evaluated by their:

Adjusted Rand Index – Similarity of two assignments (submitted cluster labels vs. ground truth), which is invariant to permutations and normalised to chance.

Homogeneity – Purity of ground truth labels within cluster.

A rank per team is established for each measurement, separately. The mean rank of both metric ranks is then the team’s final rank, which will determine the winner.

Submission

A participation in mTOP requires the submitting team to write a short paper of 4 pages (LNCS format). This will be reviewed by the organisers w.r.t.: - Clear and detailed description of applied methods - Listing of all parameter values and employed frameworks - Description of data set must not be included - Motivation for chosen method is appreciated

Please send a PDF (with standard fonts) until 12th August 2016 – 23:59 GMT with subject: "mTOP2016" and your team name. Final results will be published on the webpage.

On-Site Presentation

Talk: Three selected methods will be invited to give an oral presentation (10mins + questions). Selected teams will be announced directly after the evaluation round.

Poster: Each participating team is invited to present their work as a poster

Acknowledgment

The data set is provided by the Division of Anaesthesia at Addenbrookes Hospital, University of Cambridge.

Please reference the SMIR - The Virtual Skeleton Database in your work as:

View online | plain text | as END | as RIS | as BibTex | Mendeley | CiteULike

Download

Training Data

Please register to access the download section