Babak Esmaeili is a PhD student in the Machine Learning program at Northeastern University, advised by Jan-Willem van de Meent. Babak’s research is focused on Bayesian methods for inference in probabilistic models, probabilistic programming, and efficient inference techniques. Prior to joining Northeastern, Babak earned his bachelor’s and master’s degrees from the University of Edinburgh.
I completed the Master’s of Data Science program at the University of Edinburgh. I also completed a four-year degree in Artificial Intelligence and Computer Science at the same university.
I am interested in researching Bayesian methods for inference in probabilistic models, which provide a natural and elegant framework for modelling uncertainty. I am particularly interested in probabilistic programming, which provides exciting opportunities for abstracting probabilistic models and making the inference process easier. I am also interested in efficient inference techniques that can be applied to models involving both neural networks and graphical models.
One topic I’m currently working on is combining deep learning with probabilistic methods via variational autoencoders. Probabilistic methods are interpretable, but inference can be difficult when the model gets too complex. Furthermore, they are not particularly good at handling high dimensions. Deep learning methods, on the other hand, are very powerful at handling high dimensions and they outperform probabilistic methods with sufficient data. But they are not interpretable. An interesting research question is how we can formalize models that combine these methods using probabilistic programs and also come up with efficient inference techniques for them.
There is a lot of data out there in many fields: Heath care, chemistry, biology, vision, … With the recent developments in machine learning, many interesting patterns and models can be learned from all this data that can help the researchers in their field to have a better understating of their problems. I find applying these techniques to real-life applications quite fascinating as it enables one to explicitly observe how an adjustment in parameters or model complexity can shift the result in the system’s performance or class prediction accuracy.
Ultimately, I would like to pursue an academic career and eventually establish myself as an independent researcher in the machine learning field.
Babak Esmaeili is a PhD student in the Machine Learning program at Northeastern University, advised by Jan-Willem van de Meent. Babak’s research is focused on Bayesian methods for inference in probabilistic models, probabilistic programming, and efficient inference techniques. Prior to joining Northeastern, Babak earned his bachelor’s and master’s degrees from the University of Edinburgh.
I completed the Master’s of Data Science program at the University of Edinburgh. I also completed a four-year degree in Artificial Intelligence and Computer Science at the same university.
I am interested in researching Bayesian methods for inference in probabilistic models, which provide a natural and elegant framework for modelling uncertainty. I am particularly interested in probabilistic programming, which provides exciting opportunities for abstracting probabilistic models and making the inference process easier. I am also interested in efficient inference techniques that can be applied to models involving both neural networks and graphical models.
One topic I’m currently working on is combining deep learning with probabilistic methods via variational autoencoders. Probabilistic methods are interpretable, but inference can be difficult when the model gets too complex. Furthermore, they are not particularly good at handling high dimensions. Deep learning methods, on the other hand, are very powerful at handling high dimensions and they outperform probabilistic methods with sufficient data. But they are not interpretable. An interesting research question is how we can formalize models that combine these methods using probabilistic programs and also come up with efficient inference techniques for them.
There is a lot of data out there in many fields: Heath care, chemistry, biology, vision, … With the recent developments in machine learning, many interesting patterns and models can be learned from all this data that can help the researchers in their field to have a better understating of their problems. I find applying these techniques to real-life applications quite fascinating as it enables one to explicitly observe how an adjustment in parameters or model complexity can shift the result in the system’s performance or class prediction accuracy.
Ultimately, I would like to pursue an academic career and eventually establish myself as an independent researcher in the machine learning field.
