Today, we worked on our code for our AI model. We began to test a new model (EfficientNetV2).The new model gave us a lot of problems as the accuarcy score was very low no matter what we changed on our code. The highest our accuracy score was 24% which is considerbly low compared to our other models (EfficientNetB0 and ResNet). We began to fine tune the model to see if the accuracy would raise and while that fine tuned model was training we went into a workshop with Dr. Pramanik to learn more about skin cancer and how the different type are formed. He explained the appearance of each type of cancer and provided a checklist based on the ABCDE rule we had previously learned.He went into depth about which way each of the cancerous marks can change shape,color, and size. We also talked about each different form of the different types of skin cancer like how basal cell has a normal type and a carcinoma type which is more serious than the normal basal cell. We then looked at the difference between a cancerous skin cell and a cancerous skin cell and we identified each type of skin cancer how the different cells look. We also talked about how doctors go about testing and diagnosing for skin cancer. After this session we went back to working on our model. We noticed that the accuracy raised 10% to 34% which is better but still not good. We then began to comb through our code to see how we can make the model better. Afterwards, we took a brain break and then Blessing had us take a course on Python to earn a certification so she could go through our colab notebook and try to fix the model.
