2019-01389 - PhD Position F/M Distributed Machine Learning for IoT applications

Contract type : Public service fixed-term contract or Civil Servants Mobility (EU)

Level of qualifications required : Graduate degree or equivalent

Fonction : PhD Position

About the research centre or Inria department

The Inria Sophia Antipolis - Méditerranée center counts 37 research teams and 9 support departments. The center's staff (about 600 people including 400 Inria employees) is composed of scientists of different nationalities (250 foreigners of 50 nationalities), engineers, technicians and administrators. 1/3 of the staff are civil servants, the others are contractual. The majority of the research teams at the center are located in Sophia Antipolis and Nice in the Alpes-Maritimes. Six teams are based in Montpellier and a team is hosted by the computer science department of the University of Bologna in Italy. The Center is a member of the University and Institution Community (ComUE) "Université Côte d'Azur (UCA)".

Context

The PhD thesis is funded by Accenture Labs (industrial thesis, CIFRE)

At Inria the student will be supervised by Giovanni Neglia (NEO team) and Marco Lorenzi (Epione team).

Assignment

A complete description of the position and of the application process is at

http://www-sop.inria.fr/members/Giovanni.Neglia/19phd_federated_learning_iot_inria_accenture.pdf

Main activities

IoT applications will become one of the main sources to train data-greedy machine learning models. Until now, IoT applications were mostly about collecting data from the physical world and sending them to the Cloud. Google’s federated learning already enables mobile phones, or other devices with limited computing capabilities, to collaboratively learn a machine learning model while keeping all training data locally, decoupling the ability to do machine learning from the need to store the data in the cloud. While Google envisions only users’ devices, it is possible that part of the computation is executed at other intermediate elements in the network. This new paradigm is sometimes referred to as Edge Computing or Fog Computing. Model training as well as serving (provide machine learning predictions) are going to be distributed between IoT devices, cloud services, and other intermediate computing elements like servers close to base stations as envisaged by the Multi-Access Edge Computing framework. This approach provides at least three benefits.

  1. Reduce network load. According to recent estimates, there are 7 billions IoT devices deployed in the world. This number should increase by a factor 3 by 2025. Routing the raw data traffic generated by these devices to a few data-centers will not be feasible. It is required to extract relevant features as close as possible to the locations where data is generated.
  2. Reduce latency. ML models will be used by IoT devices to take actions in the physical world. Future wireless services for connected and autonomous cars, industrial robotics, mobile gaming, augmented and virtual reality have strict latency requirements, often below 10 ms and below 1ms for what is now called the tactile internet. A key element to satisfy such constraints is to run these services closer to the user, directly on IoT devices. Edge computing also ensures that applications are not disrupted in case of limited or intermittent network connectivity.
  3. Preserve privacy. Data captured by IoT devices can contain sensitive or private information. Pre-processing at the edge can make sure that sensitive information is removed or aggregated with data from other devices to preserve user's profile.

 

Research objectives:

The goal of this PhD thesis is to investigate how both learning tasks and prediction services can be effectively distributed across different elements in the network, taking into account computation/communication constraints.

 

During the project the candidate will:

  • Design new distributed learning algorithms with a particular focus on communication constraints;
  • Study analytically their performance and guarantees;
  • Implement distributed machine learning algorithms using libraries like PyTorch;
  • Participate to the activity of Accenture Labs, interact with research and engineering personnel;
  • Interact with Inria students and researchers, and participate to teams’ scientific.

Skills

Competences in probability, statistics, optimization, and mathematical modeling are essential (Master level). Solid programming and IT skills are necessary (Python, bash, version control systems), along with strong communication abilities.

Benefits package

  • Subsidized meals
  • Partial reimbursement of public transport costs
  • Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
  • Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
  • Professional equipment available (videoconferencing, loan of computer equipment, etc.)
  • Social, cultural and sports events and activities
  • Access to vocational training
  • Social security coverage

Remuneration

Location: Sophia Antipolis, France
Gross Salary per month: 1982€ brut per month (year 1 & 2) and 2085€ brut/month (year 3)