Methods Journal Club

The Methods JC is a journal club focused on the discussion of new methods proposed in computer vision and machine learning conferences and journals. The journal club aims to understand and discuss novel approaches proposed by the ML community, disregarding their results when out of the scope of medical imaging projects. Therefore, theoretical approaches from ML conferences without an obvious clinical impact are still interesting for the discussion. These approaches can potentially be adapted to be used in the medical imaging computing field.

The Methods JC can also be used to brainstorm new ideas or to find collaborations with researchers within the group, who can be interested or experienced in working on the proposed approach.

---

Collaborating with the JC

The topics for discussion can be proposed in several ways:

1. By coming to a meeting and proposing a topic for the next session;
2. Using the slack channel: #methods_jc kcl-bmeis.slack.com/methods_jc
3. Raising an issue in the Github repository: https://github.com/KCL-BMEIS/Methods_JournalClub/

Topics of interest

Any method/topic related to machine learning and computer vision are very welcome:

• Image registration/segmentation;
• Variational Inference;
• Kernel Methods;
• Multitask and Transfer Learning;
• Semi-supervised learning;
• Missing Data;
• Boosting and Ensemble Methods;
• Adversarial Learning;
• Reinforcement Learning;
• Stochastic Methods;
• Uncertainty Estimation;
• Unsupervised Learning;
• Domain Adaptation;
• And many others...

Schedule

The meetings happen once per month, on Thursdays 4 pm. 9th Floor, Becket House, Lambeth North

The day and room are usually announced in the slack channel.