Hey there, fellow machine learning enthusiasts! I’m currently working on a project that involves reaching an impressive 85-90% validation accuracy on the FER+ dataset using shallow neural networks. I’ve been stuck at around 70% for a while now, and I’m eager to break through that barrier. Has anyone out there achieved similar results or have any tips on how to get there?
I’ve tried tweaking my model architecture, experimenting with different hyperparameters, and even using transfer learning, but nothing seems to be giving me the boost I need. I’d love to hear from anyone who’s had success with this dataset or has any insights on how to improve my results.
Maybe we can even collaborate and share our findings to help each other out. Let me know in the comments if you’re interested!