Science Foundation Ireland (SFI) agreed to fund several Centres for Research Training (CRT), and I am involved in two of them (D-REAL and CRT-AI). D-REAL and CRT-AI seek to hire more than 240 Ph.D. students in the next four years, i.e. around 60 per year (30 in each CRT). The CRTs are unique as such that you will not only do a ‘normal’ PhD but closely work with an industry partner.
I am offering two projects in the CRTs, though these projects are not set in stone. If you have your own idea or think of some adjustments, we are open to suggestions.
Deep Meta-Learning for Automated Algorithm-Selection in Information Retrieval
The Automated Machine Learning (AutoML) community has made great advances in automating the algorithm selection and configuration process in machine learning. However, the “algorithm-selection problem” exists in almost every discipline, be it natural language processing, information retrieval, or recommender systems. Our goal is to improve algorithm selection in information retrieval and recommender systems through AutoML techniques such as meta-learning. The idea behind meta-learning is to use machine learning / deep learning to learn from large amounts of historical data on how algorithms will perform in certain scenarios. Your work might be integrated into our real-world recommender-system as-a-service that delivers millions of recommendations to our partners. You will work with the latest state-of-the-art AutoML techniques including Bayesian Optimization and Deep Siamese Neural Networks. This Ph.D. project will additionally be co-supervised by Prof Gareth Jones at DCU.
Federated Meta-Learning: Democratizing Algorithm Selection, Meta-Learning, and Automated Machine Learning
“Federated Meta-Learning” is a novel concept that allows everyone to benefit from the data that is generated through (automated) machine learning libraries including TensorFlow, scikit-learn, PyTorch, ML-Plan and (Auto)Weka. We envision a peer-to-peer or client-server architecture that allows the exchange of meta-data and models for the purpose of meta-learned algorithm selection and configuration across disciplines. The input to Federated Meta-Learning is a description of the task, and the output is a recommendation for the potentially best performing algorithm(s) to solve that task. This recommendation could consist simply of a list of the best algorithms, or their predicted performance values. The list could also consist of multiple sub-lists created with different meta learners. In its simplest form, federated meta-learning is a knowledge base or directory of algorithms-data performance measures. Ultimately, Federated Meta-Learning will be able to predict algorithm performance for unseen tasks. It will be your task to develop the first prototype of Federated Meta-Learning. Your project will include hands-on work with the latest state-of-the-art machine learning, deep learning, and automated machine learning libraries. In addition, you will develop and work on data format standards to exchange interoperable task descriptions and an API. Read more details in our proposal.
Besides our two projects, there are plenty of other interesting Ph.D. projects in the CRTs, focusing on artificial intelligence, machine learning, automated machine learning (AutoML), meta-learning, algorithm selection, recommender systems (RecSys), natural language processing (NLP), information retrieval (IR), augmented reality (AR), virtual reality (VR), and many topics more.
The Ph.D. stipends are placed at Ireland’s premier universities – in my case at Trinity College Dublin (TCD), but there are also supervisors at Dublin City University (DCU), University College Cork (UCC), NUI, Galway, University of Limerick, University College Dublin (UCD), and more.
Dr Joeran Beel
Assistant Professor in Intelligent Systems
Trinity College Dublin
School of Computer Science & Statistics
Dublin 2, Ireland
Office: O'Reilly Institute, G.15