Journal of Open Source Education

5 papers9 directions
The Journal of Open Source Education (JOSE) is a publication venue active between 2019 and 2024 that focuses on the intersection of open-source software, educational methodology, and computational research. In the context of machine learning and artificial intelligence, the venue features work that emphasizes the development of interpretable models and the application of machine learning to the physical sciences. Research published here often explores symbolic regression for scientific discovery, the stability of hybrid physics-machine learning models, and the resolution of subgrid scale physics in global climate models. The venue also serves as a platform for research into machine learning transparency and explainability. This includes the development of modular architectures for bespoke model explanations, the measurement of faithfulness in surrogate models, and the creation of tools that ground black-box decisions in human domain knowledge. Additionally, JOSE publishes work on scalable educational frameworks and software tools, such as the Neuromatch Academy for computational neuroscience and the nbgrader system for Jupyter Notebooks. Prolific contributors to the venue include researchers such as Aakash Sane, Alex Connolly, and Alistair Adcroft, who contribute to the integration of complex algorithmic components with accessible educational and visualization tools.

Papers by Year

2024 1
2022 2
2020 1
2019 1

Top Research Directions

Research directions most represented by papers from this venue.

Neural Subgrid Scale Modeling for Hybrid Physical Simulations 1
The Stability Gap in Atomistic Machine Learning 1
Hybrid Human-Algorithmic Evaluation Frameworks 1
Distilling Neural Networks into Closed Form Expressions 1
Decomposing Complex Reasoning for Granular Error Diagnosis 1
Learning Subgrid Physics for Accurate Climate Prediction 1
Explaining Black Box Models with Local Surrogates 1
The Dynamics of Linguistic Isolation in Global Science 1
Making Neural Networks Physically Interpretable 1

Top Authors

AuthorPapers
Ari S Benjamin1
Carlos Fernandez-Granda1
Cristina Savin1
Joan Bruna1
Konrad Kording1
Krishna V. Shenoy1
Marcelo G Mattar1
N Alex Cayco Gajic1
Raul Santos-Rodriguez1
Richard Gao1
Saeed Salehi1
Sara Shamekh1

Papers

TitleYear
Learning Machine Learning with Lorenz-962024
Neuromatch Academy: a 3-week, online summer school in computational neuroscience2022
What and How of Machine Learning Transparency:Building Bespoke Explainability Tools withInteroperable Algorithmic Components2022
EduCortex: browser-based 3D brain visualization of fMRI meta-analysis maps2020
nbgrader: A Tool for Creating and Grading Assignments in the Jupyter Notebook2019