Post doc - Embedded AI inspired by neuronal structures in insects

about the role

Your role is to carry out postdoc work on embedded AI inspired by insect neural structures as part of the ANR project " From biological neural connectomes to embodied ultra-edge AI " (AAPG2023) which started in 2024.

Global context and subject :

The proliferation of connected (sensors, cameras, ...) and mobile equipment, energy (, CSR), ethical () or technical () issues are driving the need to embed AI models (Edge AI) and design models that are frugal in energy and computational complexity.

Although they have small neural systems (connectome), insects can perform complex tasks (learning, calculations, action decisions, etc.

with very low energy balances and very little training data.

The connectome of certain insects is now known in detail. For example, the complete connectome of the Drosophila larva, comprising 3016 neurons and 548,000 synapses, is now known in detail.

Sub-circuits of this connectome involved in for instance in decision or classification tasks have also been characterized.

The aim of the ANR project is to take advantage of these recent advances in biology to design compact, energy-efficient artificial neural networks directly inspired by the structures and operating principles discovered in the insect connectome.

The project will focus on the development and implementation of SNNs for time series classification and prediction.

Scientific objective :

One of the main hurdles to overcome is the formalization of a methodology enabling the transfer of biological knowledge and results to the creation of DNN and SNN neural networks.

You will carry out this formalization work in close collaboration with Institut Pasteur teams. You will then apply this methodology to the design and implementation of bio-inspired DNN and SNN neural networks, and demonstrate the relevance of the approach through in-depth tests of their performance (computational and energetic).

about you

Skills (scientific and technical) and personal qualities required by the position

• Machine Learning, AI and neural networks.
• Mastery of development languages (Python, C++)
• Curiosity, dynamism and appetite for innovation and multi-disciplinary research
• Interest for biology and neurosciences.
• Excellent analytical and synthesis skills.

Required Education :

PhD in computer science, physics, mathematics or in neurosciences with a specialization in machine learning, artificial intelligence, and neural networks.

The PhD must have been obtained within the last 3 years.

Desired experience

• Spiking neural networks
• Knowledge in biology, neurosciences, biomathematics

additional information

You will be part of the Homelan / AISH team at Orange Labs Sophia Antipolis, at the forefront of expertise in AI services for the connected home (embedded and federated AI, connected equipment, home gateway, etc.

and will be part of an ecosystem of researchers, data scientists and design engineers enabling the concrete implementation of results.

• This postdoc is part of an ANR project that began in January 2024 and will run for 3 1 / 2 years. It will give rise to a major collaboration between the Homelan / AISH team at Orange Labs Sophia Antipolis and two teams of Institut Pasteur highly specialized in the analysis and modeling of insect neuronal structures.
• The subject is highly innovative. It is based on recent results in biology and neuroscience, enabling an innovative approach to the design of compact, energy-efficient neural spiking networks.
• 1-year postdoctoral fellowship, renewable twice

department

Orange Innovation brings together the research and innovation activities and expertise of the Group's entities and countries.

We work every day to ensure that Orange is recognized as an innovative operator by its customers and we create value for the Group and the Brand in each of our projects.

With 740 researchers, thousands of marketers, developers, designers and data analysts, it is the expertise of our 6,000 employees that fuels this ambition every day.

Orange Innovation anticipates technological breakthroughs and supports the Group's countries and entities in making the best technological choices to meet the needs of our consumer and business customers.

You will work in the HomeLan Services department (approx. 160 people) of the Home Services division of Orange Innovation.

This department is made up of engineers, researchers, PhD students and technical experts, and is responsible for research, anticipation and development for the Orange Group, in all areas related to the connected home, home networks, energy efficiency and savings, device management and artificial intelligence, with the aim of offering products and services at the cutting edge of innovation

contract

Post Doc