AI Training Project (Binary Classification) This project implements a binary classification AI using a self-developed training algorithm, including a custom gradient descent algorithm.

Training Code (training.py) Contains the model and kernel class definitions. The “apply” function in training.py has a variable called thres (threshold). To maximize accuracy, set the thres value based on trained models. The scale of the model can be changed in the "init" function in class "model". Details can be found in the comments in the code

Using Pre-Trained Models The repository includes three pre-trained models in .txt format. Their filenames follow this structure: "accuracy threshold value.txt" Example: 0.95 threshold 0.85.txt → Accuracy: 95%, Threshold: 0.85

How to Use a Pre-Trained Model: Rename the desired model file to AIdata2.txt. Run training.py. This will load and apply the selected model. Continue running training.py will train and optimize the model.