Developed and trained a feedforward back-propagation neural network in MATLAB using real-world cancer diagnosis data (699 patients), achieving 97% classification accuracy in distinguishing between benign and malignant cases.
Selected and evaluated network architecture components, including transfer functions (tansig), and analyzed performance using confusion matrices and training plots.
Constructed a back-propagation model in Excelto solve XOR problems, demonstrating algorithmic fluency and debugging proficiency.
Identified a higher rate of false positives than false negatives, showing the model often misclassified benign tumors as malignant—supporting the conclusion that neural networks simulate cognitive behavior without genuine understanding.
Applied theoretical frameworks from Hinton and Kriegel to assess whether neural networks can represent cognitive states or take a first-person perspective.
Critically evaluated philosophical arguments about neural networks and cognition, synthesizing computational results with cognitive theory.
Presented findings on neural networks’ limitations in representing human cognition, contributing to interdisciplinary discourse on AI and consciousness.
