News

CS Dissertation: Jarno Lintusaari reveals recent developments in approximate computational inference

In his doctoral thesis Jarno Lintusaari modeled the complex spread of tuberculosis with Approximate Bayesian computation (ABC) methods and he was one of the main developers of openly available software tool ELFI
Jarno_Lintusaari_Aalto_University_photo_Matti_Ahlgren

MSc Jarno Lintusaari got excited about statistical modeling and Bayesian inference already during his MSc studies at University of Helsinki. Encouraged by his MSc thesis supervisor Professor Jukka Corander, Jarno continued to do research and started doctoral studies at Professor Samuel Kaski’s group at the Department of Computer Science Aalto University, where he had been earlier as a research assistant. Under the supervision of both Samuel Kaski and Jukka Corander, he first studied the theory and concepts in Approximate Bayesian computation (ABC). ABC refers to a group of algorithms for approximate inference that use a simulator based model to perform statistical inference. Second, Jarno examined how ABC can provide a firm basis for inferring epidemiologies and transmission dynamics of tuberculosis.

Jarno’s supervisor Samuel Kaski’s research group, Probabilistic Machine Learning group, is very active in developing world-class machine learning methods. In such a lively and collaborative research environment, he and his co-authors were able to develop ELFI - Engine for Likelihood-Free Inference. ELFI is a Python software library for performing likelihood-free inference and a joint effort of researchers of Aalto University, University of Helsinki and University of Edinburgh. Whilst ELFI is an openly available tool, it facilitates the spread of the novel and effective statistical methods to a wider research community. ELFI provides multiple practical features for performing ABC inferences, and it can be seen as a solution for many complex research problems in which traditionally used statistical methods are not enough. Thus, ELFI will likely attract new users from researchers working with large data sets at diverse fields of research, such as in epidemiology, evolutionary biology or genetics. “Collaboration was the key in the development work of ELFI”, Jarno says. “The start of my doctoral studies was more like lonely work, but ELFI project taught me how important successful collaboration is in research.”

While Jarno’s research work in terms of doctoral dissertation is becoming to an end, he has stepped into a new career path as Data Scientist in Digital Goodie Ltd. “My new job doesn’t directly relate to my doctoral thesis, but I find my research background and skills learned during the past years very useful.” Finalizing a doctoral degree in 4 years is a tough journey and may sometimes require some creativity and organizational skills as well. “When it comes to my doctoral studies, the greatest lessons learned have been in-depth knowledge on machine learning but also skills on how to prioritize my time and work.”

MSc Jarno Lintusaari will defend his doctoral dissertation "Steps forward in approximate computational inferenceon Monday 18 March 2019 at 12 noon at the Aalto University School of Science, lecture hall H304, Otakaari 1, Espoo.

The electronic version of the thesis can be downloaded at Aaltodoc.

Contact information: MSc Jarno Lintusaari, [email protected]

  • Published:
  • Updated:
Share
URL copied!

Read more news

Opiskelijoita kävelemässä vihreiden puiden alla
Studies Published:

Donations make it possible for all the students to participate in student exchanges

International student exchange is an essential part of the studies of many students at the School of Business
Picture of Marika Tervahartiala and Kamyar Hasanzadeh after the event in front of the Hybrid Stage area.
Research & Art Published:

Perspectives on open data: Aalto Research Data Uncovered

Marika Tervahartiala and Kamyar Hasanzadeh took the stage to share their insights on opening geographic and visual data.
The Future-led Learning Podcast is hosted by Riikka Evans. Photograph: Janne Illman.
University Published:

"I'm more corageous." - Why co-teaching. Listen to a podcast

New season of Future-led learning podcast takes you to a very practical level of working with students. This season includes topical issues such as academic advising, inclusion, co-teaching and sustainability. Topic for the last episode of this season is co-teaching.
Photo of a bolometer
Press releases Published:

Quantum scientists accurately measure power levels one trillion times lower than usual

A new device can measure power levels with unprecedented accuracy and sensitivity, offering a major step forward for quantum technologies