Once AIBMod had completed its research on the Credit Loan and Credit Card Fraud Detection models in the fourth quarter of 2023, it wanted to test its model against other applicable research.

Two papers seemed to be the most relevant:

1. Revisiting Deep Learning Models For Tabular Data, Gorishniy et al. – originally published June 2021, updated Oct 2023 (“Paper 1“)
2. Well-tuned Simple Nets Excel on Tabular Datasets, Kadra et al. – originally published June 2021, updated Nov 2021 (“Paper 2“)

For reasons set out here and here and below, it was simpler to ensure like-for-like analysis with Paper 1 so that formed the the focus of the comparison runs.

AIBMod started going through the datasets in the order presented in Table 1 of Paper 1 but, in the interests of time, after running the first five datasets AIBMod became a bit more selective and picked only some of the remaining datasets.

It was more complicated to ensure like-for-like analysis with Paper 2 and details had to be confirmed with one of the members of the research team who was very helpful and replied to all queries.

Once AIBMod had completed its analysis on Paper 1 it then turned its attention to Paper 2 and, again in the interests of time, just focused on any datasets that overlapped with AIBMod’s runs from Paper 1 – that amounted to just two datasets.

For comparison results for Paper 1 please see here and here.

For comparison results for Paper 2 please see here.