We present a hybrid neural network model to solve a place recognition problem. The front end is a self-organizing net equivalent to a principal component analyzer; the back end is a feed-forward net with backpropagation, i.e. supervised learning. A conndence level greater than 0.9 was reported as the net correctly recognized a repertoire of pictures it had not seen before.

Deep learning (DL) based semantic segmentation methods have been providing state-of-the-art performance in the last few years. More specifically, these techniques have been successfully applied to medical image classification, segmentation, and detection tasks. One deep learning technique, U-Net, has become one of the most popular for these applications. In this paper, we propose a Recurrent Co...

This paper studies the problem of learning data automata (DA), a recently introduced model for defining languages of data words which are finite sequences of pairs of letters from a finite and, respectively, infinite alphabet. The model of DA is closely related to general Petri nets, for which no active learning algorithms have been introduced so far. This paper defines transparent data automat...

We introduce a new large-scale music dataset, MusicNet, to serve as a source of supervision and evaluation of machine learning methods for music research. MusicNet consists of hundreds of freely-licensed classical music recordings by 10 composers, written for 11 instruments, together with instrument/note annotations resulting in over 1 million temporal labels on 34 hours of chamber music perfor...

Structures in the solar corona are main drivers of space weather processes that might directly or indirectly affect Earth. Thanks to most recent space-based observatories, with capabilities acquire high-resolution images continuously, structures can be monitored over years a time resolution minutes. For this purpose, we have developed method for automatic segmentation observed EUV spectrum is b...

Local feature extraction remains an active research area due to the advances in fields such as SLAM, 3D reconstructions, or AR applications. The success these applications relies on performance of detector, descriptor, and its matching process. While trend detector-descriptor interaction most methods is based unifying two into a single network, we propose alternative approach that treats both c...

Much focus has been given to the Net Generation and how they learn and interact with the world. In our Centre we faced the interesting challenge of working with Net Generation graduates, and inducting them into the world of work in the emerging field of educational technology. This paper reports on a three year intern program and describes experiences and challenges of learning about e-learning.

The traditional college student of today is part of the Net Generation who has been raised in an era of instant access. Their communication and learning is complemented by the Internet, a major influence on this cohort. The regular method of contact is text messaging, instant messaging and cell phones. Learning methods for the Net Generation include Internet tools such as Web-CT, Blackboard, on...