RESEARCH FIELDNeurosciences › Other
RESEARCHER PROFILEFirst Stage Researcher (R1)
APPLICATION DEADLINE26/02/2021 23:00 - Europe/Brussels
LOCATIONFrance › Paris
TYPE OF CONTRACTTemporary
HOURS PER WEEK35
OFFER STARTING DATE01/09/2021
EU RESEARCH FRAMEWORK PROGRAMMEH2020 / Marie Skłodowska-Curie Actions COFUND
MARIE CURIE GRANT AGREEMENT NUMBER945304
“Artificial intelligence for the Sciences” (AI4theSciences) is an innovative, interdisciplinary and intersectoral PhD programme, led by Université Paris Sciences et Lettres (PSL) and co-funded by the European Commission. Supported by the European innovation and research programme Horizon 2020-Marie Sklodowska-Curie Actions, AI4theSciences is uniquely shaped to train a new generation of researchers at the highest academic level in their main discipline (Physics, Engineering, Biology, Human and Social Sciences) and master the latest technologies in Artificial Intelligence and Machine Learning which apply in their own field.
26 doctoral students will join the PSL university's doctoral schools in 2 academic cohorts to carry out work on subjects suggested and defined by PSL's scientific community. The 2020 call will offer up to 15 PhD positions on 24 PhD research projects. The candidates will be recruited through HR processes of high standard, based on transparency, equal opportunities and excellence.
Description of the PhD subject: “A computational and artificial Intelligence approach for studying dolphin communication”
Context - Motivation
Dolphins use vocal signals to communicate via two different types of sounds: pure tones and pulsed sounds. Pure tones are frequency modulated sounds such as whistles, chirps and screams. Whistles are the most studied signals in dolphins’ communication. Indeed, dolphins produce whistles in social contexts. Individual dolphins have their own “signature whistle” that are produced by other members of the group to call specific dolphins when they are out of sight and maintain group cohesion (e.g. the mother uses to call the calf, Caldwell et al., 1990). Remarkably, dolphins are the only animals other than humans capable of transmitting identity information independent of the caller's voice or location (JanikVM, 2006). Moreover, dolphins can also transmit information about food indicating the quantity and the level of hungriness of the dolphins searching for food (King and Janik, 2015). Dolphins also produce clicks signals (pulsed sounds generated at regular brief intervals containing a large spectrum of frequencies), which are used for echolocation. These clicks can also be generated at high frequencies and short intervals and are emitted within emotional or violent contexts. Despite these advances, there is still no clear evidence that dolphins use language, a method for communication consisting of meaningful sounds or symbols in a structured manner following specific grammatical rules (syntax).
Here, we propose to address this open question using data mining and artificial intelligence approaches in a unique environment (Dolphin reef, Israel, Perelberg et al. 2010). Dolphin reef is a touristic private company that promotes human-dolphin interactions. The company has 4 dolphins that are free to swim in the open sea, but during the day they prefer to rest near the reef to profit of interactions with humans. This environment combines natural behaviours with the possibility of observing and recording dolphin communication for long periods of time of 4 dolphins from which the social history is known, and individual identification and vocalizations assignment is possible.
Scientific Objectives, Methodology & Expected results
Aim 1. Combination of different sound types as “sentences” or grammatical units
Previous studies have focused on individual sound types. However, dolphins combine the different types of sounds. Here, we propose to correlate emitted sounds and the behavioural context to learn about the sounds' behavioural significance. However, in contrast to previous studies, we will use the whole richness of the entire frequency spectrogram, rather than extracting single sound types. For this purpose, we will select periods of sound emission in which the behavioural context can be assessed through the underwater cameras. Once identified, we will train convolutional neural networks (CNN) to predict the behavioural context in which the sounds were produced. One important advantage of using CNNs is that they do not need a priori assumptions or hypotheses. This approach will allow us to classify sounds according to the context they were emitted without assuming the emission of a whistle or a burst-pulse. Therefore, we will be able to classify individual sounds of different types but also the combination between them. Using this AI approach, we will learn about how dolphins combine different types of sounds to create “sentences” rather than single concepts. For example, combining a signature whistle with a specific type of burst may indicate that a given dolphin is upset or is interested in sexual intercourse. From preliminary data, we observed a specific type of whistle that was generated in a stress context (e.g. alarm whistle) and burst-pulses that could be generated alone or in combination with signature whistles. The latter were sometimes observed within the context of a fight.
Aim 2. Using vocal interactions to study syntax of dolphins acoustic communication
In linguistics, syntax is the set of rules defining the structure of sentences in a given language (e.g. the order of words). The syntax of an unknown language can be learned by studying the probabilities in the interactions between words. For example, the probability that in English language, a subject will be followed by a verb is very high. If dolphins use a language to communicate, this should have syntactic rules. To learn these rules, we will generate a large data base corresponding to the different sound categories (complex sounds, Aim 1) and whistle categories and their timestamps (the time at which they occurred). We will then use Markov chain models to reveal the grammatical structure. Markov chain models describe the probability of transitions between the different categories of whistles. For example, we may find whistles or specific types of modulations that may function as general grammatical pro-sentences (e.g. yes or no). Alternatively, the de Polavieja lab will implement AI techniques that have been shown to be extremely successful in NLP (Natural Language Processing) of the human language. One of the first successful models was word2vec. Word2Vec is a computationally-efficient predictive model for learning word embeddings from raw text. Word2vec is based on a two-layer neural network which is trained to build a low dimensional representation of words (an embedding), similarly to autoencoders. In this embedding, the distance between words indicates semantic similarity. Instead of word2vec, we will show more modern techniques, specifically Transformers, as in the BERT model. We will also study different tokenizations as different ways to give inputs to the Transformers.
As a specific test of language, we will attempt to show a long-range structure in dolphin communication. We will train BERT with masking of tokens, though this analysis could be applied to other Transformers as well. Then, we will demonstrate how the internal network structure learns to recover the possibly rich long-term structure of dolphin language. More specifically, we will use a method known as structural probe (see for an intuitive explanation). This method is able to recover syntactic trees from BERT when used on human text. We will apply it to the tokenized dolphin sound sentences and test of whether a significant syntactic structure emerges. Overall, this novel approach to study dolphin communication will allow us to shed light on rules and principles underlying the structure of dolphin acoustic communication. The existence of such a structure will suggest that their communication has syntax and therefore that dolphins use language to communicate.
The project will be supervised by two Pis (France and Portugal), but it will involve a total of 3 labs and a private company from 3 different countries (France, Portugal and Israel). The student will spend time in Israel to collect data and learn about the behaviour of each of the 4 dolphins, their routines and contexts in which the vocalizations were emitted. The collected data will be first analysed in the Sumbre lab in Paris (France) to extract the emitted sound from the background, classify them using datamining approaches and perform Markov chain models. At a second stage, the student will work in Lisbon (Portugal) in the de Polavieja lab to apply the AI approaches.
The Sumbre lab at the Ecole Normale Supérieure - PSL has 10 years of experience in analyzing high-dimensional datasets of brain dynamics (Romano et al., 2015 and Ponce et al. 2018) and many more on the analysis of behaviour. The de Polavieja Lab at the Champalimaud Institute (Portugal) has 30 years of experience in Applied Mathematics, and in the last 5 years began to develop and apply AI methods for Biology. More specifically, he has used Machine Vision and attention models for analysis and modeling of animal behavior (Heras et al., 2019) and (Romero-Ferrero et al. 2019). The collaboration also includes the Shashar lab (University of Beersheva, and head of the Dolphinlab at the Dolphin reef, Israel, expert in marine animal behaviour and communication) and the Dolphin reef (Israel, a private company promoting human-dolphin interactions and free dolphin therapies for children with disabilities). Although they will not be involved in data analysis, they are fundamental for data acquisition and advice on dolphin behaviour.
Germán Sumbre and Gonzalo de Polavieja
Created in 2012, Université PSL is aiming at developing interdisciplinary training programmes and science projects of excellence within its members. Its 140 laboratories and 2,900 researchers carry out high-level disciplinary research, both fundamental and applied, fostering a strong interdisciplinary approach. The scope of Université PSL covers all areas of knowledge and creation (Sciences, Humanities and Social Science, Engineering, the Arts). Its eleven component schools gather 17,000 students and have won more than 200 ERC. PSL has been ranked 36th in the 2020 Shanghai ranking (ARWU).
- Opportunity to conduct academic research in a top 100 university in the world.
- High-quality doctoral training rewarded by a PhD degree, prepared within Ecole Normale Supérieure - PSL and delivered by PSL.
- Access to cutting-edge infrastructures for research & innovation.
- Appointment for a period of 36 months (job contract delivered by the involved component school of PSL) based on a salary of 3100 € gross employer (including employer tax) per month or approximately a 2228 € gross salary per month.
- Job contract under the French labour legislation in force, respecting health and safety, and social security: 35 hours per week contract, 25 days of annual leave per year (“congés annuels”). Eventual complementary activities may be accepted or proposed by the co-supervisors (maximum of 64h/year for teaching, 32 day/year for specific missions).
- Short stay(s) or secondment in France or abroad are expected.
- An international environment supported by the adherence to the European Charter & Code.
- Access to AI training package, with a strong interdisciplinary focus, together with a Career development Plan.
- Applicants must have a Master’s degree (or be in the process of obtaining one) or have a University degree equivalent to a European Master’s (5-year duration) to be eligible at the time of the deadline of the relative call.
- There is no nationality or age criteria, but applicants must not have resided or carried out their main activity (work, studies, etc.) in France for more than 12 months in the 3 years immediately before the deadline of the call (MSCA Mobility rule).
- Applicants must declare to be available to start the programme on schedule.
For submitting your online application, go to: https://www.psl.eu/recherche/grands-projets-de-recherche/projets-europee...
The online application should contain the following documents:
- English translated transcripts from the Master’s degree (or equivalent 5-year degree). A copy of the Master’s degree or a certificate of achievement will be required later on for the final registration.
- International curriculum vitae and a cover letter explaining the reasons that lead him/her to prepare a PhD, why he/she applies to this offer and his/her professional project (guidelines will be given to the applicants in order to help him/her in the writing of his/her letter).
- Two academic reference letters.
- A statement duly signed on the mobility rules, availability, and conflicts of interest.
The applicants can only apply to one PhD project among the available ones. Multiple applications of one candidate will automatically make all his/her applications ineligible.
The applications will be analysed by the Management Team for eligibility and completeness. Afterwards, the applications will be reviewed by the Selection Committee. In the pre-selection round (March-April 2021), applicants will be rated using a scoring system based on 3 criteria (academic excellence, experience, motivation, and qualities). A shortlist of qualified applicants will be interviewed during the selection round (June 2021) to further assess their qualifications and skills according to the predefined selection criteria.
All information regarding the applications (criteria, composition of the Selection Committee, requirements) can be found on the website of the programme, in greater detail.
The selection and recruitment processes of the PhD student will be in accordance with the European Charter for Researchers and Code of Conduct of the Recruitment of Researchers. The recruitment process will be open, transparent, impartial, equitable, and merit based. There will be no discrimination based on race, gender, sexual orientation, religion of belief, disability, or age.
The Institut de biologie de l’Ecole Normale Supérieure (IBENS) was created in 2010. It is a public research unit involving the ENS - PSL, CNRS and INSERM. The IBENS is a fundamental research and multidisciplinary center that conducts innovative projects aimed at deciphering the mechanisms and principles leading to the function of living systems. Internationally recognized for the quality and originality of its work, IBENS hosts more than 300 people grouped in 30 autonomous teams conducting highly collaborative research that combines experimental and theoretical approaches in a strong translational manner. Since its creating, the IBENS has been successful in securing competitive national and international public funds, among them, 15 ERC grants. Beyond research, IBENS is also a place for technological innovation (patenting, hosting start-ups), and for the transmission of knowledge to the younger generations. Thus, our institute, within the Biology Department of the ENS - PSL, is strongly involved in the teaching and training of students and young researchers at all levels. Through research training through research, IBENS aims to prepare and inspire tomorrow's researchers in life sciences.
The Ecole Normale Supérieure - PSL is a leading multidisciplinary institution that focuses on training through research. The ENS - PSL defines and applies scientific and technological research policies, from a multidisciplinary and international perspective and counts close relationships with prestigious partners, in France and abroad. It encompasses fourteen teaching and research departments, spanning the main humanities, sciences, and disciplines. The ENS - PSL currently has a staff of almost 800 lecturers, ENS - PSL, CNRS or associated researchers, and post-doc researchers. Within its Departments, the ENS - PSL includes 40 research units identified as ENS - PSL, INSERM or INRIA, encompassing ENS - PSL and CNRS agents as well as 300 foreign researchers and 650 doctoral students. The ENS - PSL respects the principles of the European Charter and Code for Researchers and is engaged in the HRS4R certification.
Web site for additional job details
Required Research Experiences
RESEARCH FIELDBiological sciences
YEARS OF RESEARCH EXPERIENCE1 - 4
REQUIRED EDUCATION LEVELBiological sciences: Master Degree or equivalent
REQUIRED LANGUAGESENGLISH: Excellent
- Background both in Mathematics/Computer Science and Biology.
- Good programming skills, especially in Python.
EURAXESS offer ID: 579050
The responsibility for the jobs published on this website, including the job description, lies entirely with the publishing institutions. The application is handled uniquely by the employer, who is also fully responsible for the recruitment and selection processes.
Please contact firstname.lastname@example.org if you wish to download all jobs in XML.