data-centric-ai VS DikeDataset

In line with initiatives like Data-centric AI (https://https-deeplearning-ai.github.io/data-centric-comp/, https://github.com/HazyResearch/data-centric-ai), we firmly believe that iterating on datasets (finding label errors, dataset slicing, QA, etc.) will become more and more important, and tools for making this easier and involving different roles are needed.

[2] DikeDataset

Hi, Reddit, During the project implementation for my bachelor's thesis [1], a software (named dike, as the Greek goddess of justice) capable of analyzing malicious programs using artificial intelligence techniques, I was unable to locate an open source dataset with labeled malware samples in the public domain. As a result, I created DikeDataset, a dataset with labeled PE and OLE samples [2]. Because it was not the main focus of my thesis, the samples attributes are not evenly distributed (the benign-malicious and OLE-PE ratios are quite low), but the dataset aided greatly in the research process. This week, I was surprised to see that the public GitHub repository (which was used only for storage, without any promotion on communities like this) gained some organic reach (views, clones and stars). Furthermore, I was thrilled to learn that it was used in a research article published in 2021 [3]! As a result, I'd like to share this project with the community in the hopes that it will be useful to some members of the community. [1] dike [2] DikeDataset [3] Toward Identifying APT Malware through API System Calls