0000000000685744
AUTHOR
Giosue Lo Bosco
Question Answering with BERT: designing a 3D virtual avatar for Cultural Heritage exploration
Recent technological developments are changing how people experience physical and virtual environments. Technologies like Virtual Reality (VR), Augmented Reality (AR), and Mixed Reality (MR), today, are impacting daily life and are being used in various domains including Cultural Heritage. Applying intelligence to applications using these technologies, through AI and Deep Learning can provide a more immersive user experience. In this paper, we propose the design of a system that realize an avatar with question-answering capabilities, for the specific purpose of giving help in immersive navigation of cultural heritage sites. It is based on a BERT model tailored to the Italian language and wi…
A Case Study for the Design and Implementation of Immersive Experiences in Support of Sicilian Cultural Heritage
Virtual Reality (VR) is a robust tool for sponsoring Cultural Heritage sites. It enables immersive experiences in which the user can enjoy the cultural assets virtually, behaving as he/she would do in the real world. The covid-19 pandemic has shed light on the importance of using VR in cultural heritage, showing advantages for the users that can visit the site safely through specific devices. In this work, we present the processes that lead to the creation of an immersive app that makes explorable a famous cultural asset in Sicily, the church of SS. Crocifisso al Calvario. The application creation process will be described in each of its parts, beginning from the digital acquisition of the …
Learned Sorted Table Search and Static Indexes in Small Model Space
Machine Learning Techniques, properly combined with Data Structures, have resulted in Learned Static Indexes, innovative and powerful tools that speed-up Binary Search, with the use of additional space with respect to the table being searched into. Such space is devoted to the ML model. Although in their infancy, they are methodologically and practically important, due to the pervasiveness of Sorted Table Search procedures. In modern applications, model space is a key factor and, infact, a major open question concerning this area is to assess to whatextent one can enjoy the speed-up of Learned Indexes while using constant or nearly constant space models.We address it here by (a) introducing…
An Active Learning Approach for Classifying Explosion Quakes
In this work, an Active Learning approach for improving the classification of passed seismo-volcanic events is proposed. Here we study the specific case of Explosion Quakes from Stromboli Volcano versus other seismo-volcanic events, recorded as seismograms, and the use of Random Forest as a Classification method. In conformity with the active learning paradigm, the approach recalls the human intervention for the annotation of uncertain data. The uncertainty is established by the event probabilities, predicted by a trained random forest classifier. The human intervention consists of editing and relabelling the data into these main three classes: Explosion Quakes, Non-Explosion Quakes or Non-…
Deep Metric Learning for Histopathological Image Classification
Neural networks demonstrated to be effective in multiple classification tasks with performances that are similar to human capabilities. Notwithstanding, the viability of the application of this kind of tool in real cases passes through the possibility to interpret the provided results and let the human operator take his decision according to the information that is provided. This aspect is much more evident when the field of application is bound to people's health as for biomed-ical image classification. We propose for the classification of histopathological images a convolutional neural network that, through metric learning, learns a representation that gathers in homogeneous clusters the …