{"blocks": [{"key": "455a2671", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "1033dc68", "text": "Technical deep-dive on modelling choices used in your project and OA multiple-choice concepts.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "fe39b714", "text": "Question", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "58484aac", "text": "Explain how and why you applied PCA and L2 normalization before training. Walk through logistic regression versus alternative baselines you considered. Describe backpropagation in modern neural nets. What is knowledge-informed machine learning and how did you tune the threshold to improve both FPR and TPR?", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "ebe91e91", "text": "Hints", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "b8b43258", "text": "Connect preprocessing to variance reduction, articulate loss gradients, and discuss evaluation trade-offs.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}], "entityMap": {}}