In this work, we investigate how to improve semi-supervised DNN for low resource languages where the initial systems may have high error rate. We propose using semi-supervised MLP features for DNN training, and we also explore using confidence to improve semi-supervised cross entropy and sequence training. The work conducted in this paper was evaluated under the IARPA Babel program for the keyw...

In this paper, we introduce the concept of soft supra strongly semi∗ generalized closed sets (soft supra strongly semi∗ g-closed for short) in a supra soft topological space (X,μ,E) and study their properties in detail. The relationship between soft supra strongly semi∗ generalized closed sets and other existing soft sets have been investigated. It has been pointed out in this paper that many o...

Intonational typology has recently begun to scrutinize the typology of accent beyond the common binary distinction of ‘stress accent’ and ‘pitch accent’ (or ‘non-stress’ accent). Contributing to this research, this study considers the phonetic correlates of accent and its patterns of distribution in the Kuninjku dialect of Bininj Gun-wok (BGW), a polysynthetic language spoken in Northern Austra...

IN THIS PAPER the discovery in Japan of smaller Foraminifera from the Upper Cretaceous strata is recorded for the firsi: time. This contribution, in addition to its specific biostratigraphic application in Japan, has further interest by reason of its extension of the geo. graphic range ,of the widely distributed Cretaceous pelagic foraminiferan, Globotruncana. In the north Pacific region, Globo...

In many applications, the dearth of information to a proper training and use of supervised Machine Learning methods is a persistent problem. This fact led to the development of the semi-supervised learning paradigm. This paradigm can be understood as a combination of concepts of unsupervised and supervised paradigms. The way how the learning is conducted allows to organize the semi-supervised m...