Marcus Collier
Associate Professor Of Sustainability Science
Dublin, Dublin 1, Ireland
Summary
👤
Senior🎓
Top SchoolMarcus Collier is an Associate Professor of Sustainability Science at Trinity College Dublin with over a decade leading large EU-funded projects that translate social-ecological research into urban resilience and nature-based solutions. He has directed TURAS and Connecting Nature and now leads an ERC Consolidator-funded NovelEco citizen science project probing public attitudes to urban novel ecosystems, blending rigorous ecology with transdisciplinary co-creation. His work spans landscape ecology, urban ecology, and governance, and he pairs academic leadership (Director of Research) with hands-on innovation in project design and stakeholder engagement. Less obvious: he has contributed technical work on uncertainty and out-of-distribution evaluation in machine-learning benchmarks (Google’s uncertainty-baselines), underscoring a rare combination of environmental science and computational skills.
10 years of coding experience
17 years of employment as a software developer
CUS Leeson Street, Dublin 2PhD, Environmental Policy, PhD, Environmental Policy at University College Dublin
Github Skills (11)
machine-learning10
uncertainty10
deep-learning10
tensorflow10
estimate10
evaluation10
data-science10
metric10
probabilistic-programming8
statistics8
bayesian-methods8
Programming languages (3)
Github contributions (5)
google/uncertainty-baselines
Aug 2021 - Jan 2023
High-quality implementations of standard and SOTA methods on a variety of tasks.
Role in this project:
Data Scientist
Data ScientistContributions:34 commits in 1 year 5 months
Contributions summary:Marcus contributed to the implementation of ECE (Expected Calibration Error) and OOO (Out-of-Distribution) evaluation metrics, specifically for the heteroscedastic model within the JFT (Jointly Fine-tuned) framework. Their work involved modifying the `baselines/jft/heteroscedastic.py` file to incorporate these metrics, including the addition of CIFAR-10H and ImageNet ReaL evaluation. Furthermore, the user made changes in experiment configuration files to enable the evaluation of these metrics. These modifications aimed to enhance the evaluation of uncertainty in the model.
implementationsstatisticsdata-sciencedeep-learningneural-networks
MarkPKCollier/NeuralTuringMachine
May 2018 - Jun 2019
Tensorflow implementation of a Neural Turing Machine
Contributions:23 commits, 3 PRs, 27 pushes in 1 year 1 month
tensorflowturing-machine
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.