Johan Dahlin is a business-minded PhD who is passionate about using Bayesian data analysis to make the world a better place.
Who am I?
I began life in Enköping, Sweden in 1986. As a young boy, I quickly found an interest in Mathematics and Computer Science, which led me to study Engineering and then to pursue a PhD based on research and coursework.
I received my PhD in Automatic Control in May 2016 after successfully defending my thesis which contained a total of 17 peer-reviewed papers (13 conference papers and 3 journal papers) published at top conferences and journals in the fields of Computational Statistics and System Identification.
Currently, I am developing automated data-driven algorithms for building dynamical models as a PostDoc at the School of Engineering at the University of Newcastle, Australia.
During my research career, I have worked at a number of different companies and universities. I visited Prof. Robert Kohn at he University of New South Wales, Australia during the autumn of 2014 as part of his PhD studies. I have also worked as a Research Scientist at Sectra AB and as a PostDoc at the division of Statistics and Machine Learning at Linköping University.
Short CV summary
Some of my past projects
Teaching computers to detect cancer
Breast cancer is one of the most common forms of cancers in women with about two million new cases every year. Early detection is the best approach to improve survival. The successful application of deep learning for processing mammograms would enable more extensive screenings of the population. Such algorithms would help by sifting through the gigabytes of data collected to flag abnormalities that require further analysis by an expert.
Photo: Nevit Dilmen.
Finding survivors in disaster areas
It is often difficult to coordinate and plan the rescue work after a disaster such as an earthquake or a Tsunami. Statistical models can be a help in these situations by combining information from cameras mounted on UAVs, maps, cell tower information, etc. to create a probability map over an area. This map would tell the rescuers where it is most likely to find survivors after a disaster, so that they might focus their effort and resources to save as many people as possible.
Photo: Mark Dixon.
Finding alternative medical diagnoses
Electronic health records of patients are usually stored on a central server at hospitals. These are a possible gold-mine of information that is currently large under-utilised. However, Machine learning and Natural language processing methods can be used to extract useful information from these records. This can help medical doctors to make better diagnoses by presenting lists of possible diagnoses given certain test results and symptoms.
Photo: NEC Corporation of America.
Would you like help with exploring and analyzing a data set? Would you like to have tailored efficient algorithms for your problem? Would you like to have help with training and education? I can help!
I have many years of practical experience in developing and applying methods from Bayesian statistics, Machine learning and Artificial intelligence both in academia and industry. My PhD education has provided me with a deep understanding of these methods, which is essential when tailoring them to specific problem or training others. I have experience in working with numerical, text, image and network data. Get in touch for more information.
More about my research
My research is situated somewhere in the intersection of Computational statistics, Machine learning and System identification (the Engineering discipline of building models of dynamical systems for control). I gave some application examples on the previous slide and on this some more general interests are listen. For more details, use the buttons to navigate to my publication list and my talks.
My research interests
Accelerating Monte Carlo algorithms for Bayesian inference, such as particle filtering, sequential Monte Carlo and Markov chain Monte Carlo.
Inference in longitudinal data (short data records of many individuals) for applications within Medicine and recommendation systems.
Approximate Bayesian inference by modifying the model to simplify inference, e.g., using Gaussian process optimization.
Developing methods for applying Machine/Deep Learning for applications in Climate Science, Medicine and Finance.
Some current hobby projects
I like to keep busy during my spare time with being outdoors, reading books, running, swimming, weight-lifting, yoga, cooking and socializing. Each year, I decide on some special focus areas and the current ones are listed below.
Would you like to know more?
I love to connect with new people for networking and to share experiences and idea. Therefore, do not hesitate to get in touch with me if you have any questions connected to my research, the source code on GitHub or if you would like to discuss a business idea or hire me for a job.