During my thesis, I read a bunch of books and papers. Sometimes, we have a rough memory about one paper. This list has two purposes:

• get an idea of my research topic
• keep an history of what I have read

I may try to add a comment, and some keywords, but sometimes, no time for that.

2023

Malware detection with opcode graph analysis

Malicious URL

A survey on Malicious URL detection

White Box Cryptography

White Box Cryptography and AES implementation

Multiview-learning

A Survey on Multi-View Clustering, 2018 Too much about maths and detail. Not good as an introduction

Good intro to the problem.

Relationship between ensemble clustering and multiview clustering

three assumptions:

• Sufficiency: Each view is sufficient for the learning task on its own,
• Compatibility: The objective functions export the same predictions for co-occurring features with high probability in both views,
• Conditional independence: All views provide learning labels in conditional independence.

Co training, fusion

Multiview Clustering

Separation of clustering outcomes reduce the overall entropy

New Approaches in Multi-View Clustering, http://dx.doi.org/10.5772/intechopen.75598

Multivew clustering is kind of equivalent to datafusion. Get different inputs or views from the data (for web pages: page content, page hyperlink), and combien them together

For instance with KMeans, the features of each views do not interact (no way to compute distnace between them) However, clusters must contains the same items in each view.

Ensemble Clustering

Bayesian Clustering Ensemble

MCMC,

Model Explainability

• Forough Poursabzi-Sangdeh, Daniel G Goldstein, Jake M Hofman, Jennifer Wortman Wortman Vaughan, and Hanna Wallach. 2021. Manipulating and Measuring Model Interpretability.

User study: example on predicting appartment prices. Give clues to user, how to compute, VS a model. Because too many features, users tend to not follow advices. Explainable model doesn’t help that much.

TODO