Special Session on "Sparse Data Machine Learning for Domains in Multimedia" at CBMI 2017, June 19-21, Florence - Call for papers
This is a cordial invitation to submit a paper to the special session on "Sparse Data Machine Learning for Domains in Multimedia" at CBMI 2017- Content-Based Multimedia Indexing, to be held in Florence, Italy (19 - 21 June, 2017).
Sparse Data Machine Learning for Domains in Multimedia
Using multimedia in specific fields such as medicine or psychology is an emerging trend, which gives the multimedia community an opportunity to perform research that can have societal impact and help people. Nevertheless, these domains often come with some challenges. One of the biggest challenges is the availability of labeled data. In the area of medicine, there are some areas and diseases that are well covered, for example image and video data for polyp detection. Thus, even the well covered areas have, compared to other datasets, very little data. This fact makes it very challenging to apply machine learning methods and get meaningful results. Especially, the highly praised and used deep learning is very depending on a lot of good training data. Moreover, deep learning might not be the silver bullet for tackling every problem. With this special session, we want to emphasize that there are relevant and emerging topics that might not or are against the odds be solvable with deep learning and require a more open and broader point of view.
The scope of this session is machine learning in areas that come with a lack of data, such as medicine, addressing the challenges described in the motivation.
Examples for topic of interest are:
- Traditional machine learning vs Deep learning
- Machine Learning on small data
- Deep learning with small datasets
- Unsupervised machine learning
- Data creation
- Data annotation
- Knowledge transfer
- Areas with small multimedia datasets (for example medicine, psychology)
- Multimedia tools and applications in fields with lack of data (for example medicine, psychology)
More information about the conference can be found here.
For guidelines and submission procedure see Paper submission.
The CBMI proceedings are traditionally indexed and distributed by IEEE Xplore and ACM DL.
In addition, authors of the best papers of the conference will be invited to submit extended versions of their contributions to a special issue of Multimedia Tools and Applications journal (MTAP).
- Full/short paper submission deadline: February 28, 2017
- Notification of acceptance: April 10, 2017
- Camera-ready papers: April 21, 2017
- Konstantin Pogorelov - Simula Research Laboratory, Norway - konstantin(at)simula.no
- Duc-Tien Dang-Nguyen, Dublin City University, Ireland - (duc-tien.dang-nguyen(at)dcu.ie)
- Michael Riegler - Simula Research Laboratory, Norway - michael(at)simula.no
- Luca Piras, University of Cagliari, Italy (luca.piras(at)diee.unica.it)
- Paolo Rota - Vienna University of Technology, Austria - rota(at)caa.tuwien.ac.at
- Pål Halvorsen - Simula Research Laboratory, Norway - paalh(at)simula.no
Biometric Technologies for Computer Security Seminar, January/February 2017, registration form available.
The registration form for the Biometric Technologies for Computer Security seminar, held by Prof. Gian Luca Marcialis, is now available.
Each lesson will take 4 hours. Calendar: 2, 3, 6, 15, 16, 17 feb. 2017, h15.00-19.00 room B1DIEE (on february 3 the lesson will be held in room B0).
This series of lectures is aimed to introduce the fundamentals of biometric technologies for personal recognition.
A biometric system for personal verification is made up of the following modules: sensor, feature extractor, matcher. These modules will be described with respect to two biometrics: fingerprint and facial traits. Multi-modal biometrics, that is, systems which combine information from multiple biometrics, will be also presented. Finally, some open research issues will be introduced.
Credits: 1 (project) - 2 (final exam).
GreenBit presents a new product family and a brand new fake finger detection software developed in collaboration with PRA Lab's biometric unit
Green Bit presents a new product family called DactyID and the new “Fingerprint Anti-spoofing” software feature which has been developed in collaboration with the Electronic Engineering Department of Cagliari University, PRA Lab's biometric unit.
From the Green Bit website:
The brand new DactyID family has been designed having in mind the fast expanding market of digital transactions and added value e-services, such as: Banking, Healthcare, Commerce, etc. Furthermore the offer of reliable, robust and high quality FBI certified products will be a winning factor for a secure registration, identification, authentication of citizens in these market segments.
The DactyID20 is the first state-of-art model of this family. It is a compact single-finger (FAP 20 FBI/PIV certification in progress) livescan, offering a superior built-in patented liveness finger detection system. It is available in two versions: as desktop unit and as OEM version for its integration in third-party systems. It will offer, combined with our DactyMatch SW package, a perfect solution for a Secure Identification.
The new MultiScan_SDK add-on (the GBT_FFD-DIEE), a fake-finger detection feature, can be used on the flagship GreenBit product DactyScan84c and doesn’t require any physical upgrade or HW modification. The algorithms are capable to “learn” during use, thus improving efficiency over time.
GreenBit will be also sponsor of the 5th International Fingerprint Liveness Detection Competition (previous edition).