Feedback on the part on results

I recompiled the report myself, and I wonder if I did something wrong, because the results section is inconsistent and contains many typos. Is it a draft? Did you forget to push commits?

In its present state, it requires major changes. Some of them are:

- Section numbers are wrong (double numbers);
- The legend of the "Small Datasets: Efficiency vs. Accuracy Trade-off" figure is incomplete: what do shapes stand for?
- This legend is inconsistent with the text.
- The figure "Impact of CR on Large Datasets" makes no sense at all.
- A figure appears to be missing in the clean versus noisy analysis.
- The discussion feels like chatGPT and is not in line with the results (it mentions a heatmap that is not there any more).