14 November 2018

2nd Environmental Acoustic Big Data workshop

Learning of Biosonars or Songs & other Models to Survey Marine Mammal
This workshop aims to demonstrate the needs and realisation of long term, big data monitoring of the biodiversity, by leveraging advances in IoT, versatil microcontrollers, advanced signal processing and machine learning. Continuous monitoring of ecosystems can be achieved by recording and analysing their acoustic environment (soundscape), which is the expression of the habitat quality, richness and biodiversity. Managing big acoustic data streams requires specialized tools to classify the scenes and detect / recognise the different components, e.g. biophony, geophony and anthropophony. This second edition will be devoted to model and applications on mammals.

Proceedings are aimed to be published jointly with the first workshop. Free Registration is required before 10th nov. To register please contact herve.glotin@lis-lab.fr , subject = “EADM”


9h Welcome

9:15 Opening & objectives
« Data Science and bioacoustics of Biosonar & Songs of marine mammals »
Hervé Glotin, LIS CNRS U. Toulon

9:30 Case studies, overview
« Marine soundscapes, to broaden the scope of underwater acoustic monitoring »
Gianni Pavan, CIBRA, univ Pavia, Italy

« Acoustic monitoring in Pelagos sanctuary »
Marion Peirache, Parc national de Port-Cros & Pelagos, France

« Caribbean survey of marine mammals »
Jeffrey Bernus, Scientific resp. of CARIMAM, AGOA & AFB, Martinique

10:30 – 10:45 Coffee break

“ACCOBAMS addressing underwater noise issues »
Maÿlis Salivas, Accobams, Monaco

11:05 Emb. PAM on Autonomous Surface Vehicle completed by satellite for large scale survey

« Sphyrna : an oceanic big data scientific Autonomous Surface Vehicle »
Fabien de Varenne, Seaproven SA

« Passive acoustic 3D tracking of deep cetacean diver from the Sphyrna ASV »
Marion Poupard et al., Biosong SA

12:30 lunch – 14:00

« High Velocity acquisition : JASON sound card, 3D Tracking of Biosonar of pink Amazon River dolphins into the wild »
Valentin Barchasz et al., SMIoT

14:20 Deep Learning for detection and Classification

« Bird Species Identification from Audio Recordings and Metadata »
Jan Schlüter et al., LIS CNRS & OFAI Austria

« Towards Deepnet AutoEncoder – application to Biosonar of Marine Mammal »
Maxence Ferrari et al., DGA LIS LAMFA

« Deep learning and satellite images, perspective on whale survey »
Adeline Paiement, LIS & Bristol univ.

15:20 – 15:40 Coffee break

15:40 Detectop, Pattern analysis and inversion

« Mathematical model for long term analysis of the frequency of songs of Blue or Humpback whales »
Julie Patris & Franck Malige, AMU, STIC AmSud BRILAM, Chili (visio)

« Analysing Weeks of Underwater Sound in Hours: Preliminary Results »
Jan Schlüter et al., LIS CNRS & OFAI Austria

16:10 Discussion

17:00 Closing


Glotin, Ferrari, Marxer, Paiement

Contact / info : herve.glotin@lis-lab.fr


TGV Toulon, Airport Toulon, Campus La Garde, http://www.univ-tln.fr
14 November 2018, 09h0017h00
Classroom M001
Buiding M
University of Toulon, France

Prochains évènements

Retour à l'agenda
15 November 2018

Séminaire CaNa : 15 novembre à 10h : Adiabatic methods in quantum control, CHITTARO Francesca

Résumé : Quantum control is the branch of control theory concerned with quantum systems, that is, dynamical systems at atomic scales, that evolve according to the laws of quantum mechanics. One fundamental issue in quantum control theory is the controllability of quantum systems, that is, whether is it possible to drive a quantum system to a desired state, by means of suitably designed control fields. To cover most theoretical and practical situations, several notion of controllability have been proposed, as, for instance, controllability in the evolution operator, pure state controllability, controllability in population and eigenstate controllability. After a brief introduction on the topic, I will expose some results on the approximate spread controllability of (closed) quantum systems, obtained by means of adiabatic techniques, and taking advantage of the presence of conical intersections between the energy eigenstates. These results have been published in [1],[2]. Adiabatic techniques can be successfully used also to study the dynamics of open quantum systems. In particular, in [3] they have been applied to find an effective description of the evolution of open, weakly coupled quantum systems, where the sub-system of interest dissipate with much slower time scales than the rest of the system. [1] U. Boscain, F. C. Chittaro, P. Mason, M. Sigalotti Quantum Control via Adiabatic Theory and intersection of eigenvalues IEEE-TAC, (2012) 57, No. 8, 1970--1983 [2] F. C. Chittaro, P. Mason Approximate controllability via adiabatic techniques for the three-inputs controlled Schrödinger equation, SIAM J. Control Optim., (2017), 55(6), 4202–4226. [3]