Industrial application of Machine Learning for the telecommunication industry - QC-103

Preferred Disciplines: Computer Science, Mathematics. Post-Doc
Company: Ericsson
Project Length: 24 months
Desired start date: July 1, 2017.  
Location: Montreal, Quebec
No. of Positions: 2
Preferences: Montreal region a preference but not mandatory. No preferences for language.

About Company:

Ericsson is a world-leading provider of telecommunications equipment and services to mobile and fixed network operators. Over 1,000 networks in more than 180 countries use Ericsson equipment, and more than 40 percent of the world's mobile traffic passes through Ericsson networks. Using innovation to empower people, business and society, we are working towards the Networked Society, in which everything that can benefit from a connection will have one. At Ericsson, we apply our innovation to market-based solutions that empower people and society to help shape a more sustainable world.

Project Description:

Ericsson helped shape the world through connectivity and mobility, one of the most important technology platforms of our time. The telecommunication industry is facing major disruption as it seeks to create more intelligence inside the network and is in the process of adopting Machine Learning techniques for application to the telecommunication network (including IoT and industrial processes) analysis and control. Machine Learning techniques are also being applied to subscriber/devices/machines behavior analysis. Machine learning at Ericsson has already been successfully used, for example in order to reduce deployment time, solve customer issues pro-actively and make services more efficient. Ericsson sees four stages in the development of engineered intelligence in telecommunication networks:

Reactive: Knowing what happened (this is generally where the telecommunication industry is at today)
Predictive: Using ML to predict future outcome; I know what will happen.
Suggestive: Suggest outcomes based scenarios; I know what you would like to happen.
True intelligence: Mapping cause and effect and pro-actively solve issues before they are perceived; I can take care of myself.

Ericsson can bring together knowledge, experience and data across borders and across the planet in order to create solutions.  

Background and required skills

Research Objectives/Sub-Objectives:

Through a series of concrete telecommunication industry use cases, find ways of applying Machine Learning techniques (subscriber management, network optimization and automation, fraud detection, IoT, …) on multi-dimensional features and time series.

  • Preprocessing techniques
  • Proper feature transformation, reduction and selection
  • Model selection


  • While the specific methodology for the project can be evolved, the general expectation is  similar to what is found in existing literature for ML applications to real-world problems.

Expertise and Skills Needed:

Problems in the field require application of Classification, Regression and/or Clustering and generally require Preprocessing, Feature Selection/Dimensionality reduction and Model selection. The Data itself may include multi-dimensional features and time series. Good knowledge of those areas and associated tools (in Python and R) is a requirement.

Ability to translate problems definition from the telecommunication domain to the Machine Learning domain as well as ability to find a solution path, experiment it and communicate back the solution and approach are also a must.

For more info or to apply to this applied research position, please

  1. Check your eligibility and find more information about open projects.
  2. Complete this webform. You will be asked to upload your CV. Remember to indicate the title of the project(s) you are interested in and obtain your professor’s approval to proceed!
  3. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform or directly to Jesse Vincent-Herscovici   at, jvh(a)