embed_map.json

{
  "rules": [
    {
      "image_match": "example1.png",
      "notebook_html": "knn_example1_interactive.html",
      "notebook_ipynb": "knn_example1_interactive.ipynb",
      "section_title": "Interactive Notebook: k-NN Example 1",
      "description": "Click inside the interactive plot to move the new point. Classification updates automatically.",
      "iframe_title": "Interactive k-NN Example 1 Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "example2.png",
      "notebook_html": "knn_example2_interactive.html",
      "notebook_ipynb": "knn_example2_interactive.ipynb",
      "section_title": "Interactive Notebook: k-NN Example 2 (Outlier Sensitivity)",
      "description": "Use Manhattan distance, vary k, and toggle the outlier to see how the prediction changes.",
      "iframe_title": "Interactive k-NN Example 2 Notebook",
      "iframe_height": 780
    },
    {
      "image_match": "gen-logis.png",
      "notebook_html": "generalized_sigmoid_interactive.html",
      "notebook_ipynb": "generalized_sigmoid_interactive.ipynb",
      "section_title": "Interactive Notebook: Generalized Sigmoid Function",
      "description": "Adjust scale a, slope b, and shift x0 in a/(1+e^{-b(x-x0)}) to see how the sigmoid changes.",
      "iframe_title": "Interactive Generalized Sigmoid Function Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture4/sigmoid.png",
      "notebook_html": "generalized_sigmoid_interactive.html",
      "notebook_ipynb": "generalized_sigmoid_interactive.ipynb",
      "section_title": "Interactive Notebook: Generalized Sigmoid Function",
      "description": "Adjust scale a, slope b, and shift x0 in a/(1+e^{-b(x-x0)}) to see how the sigmoid changes.",
      "iframe_title": "Interactive Generalized Sigmoid Function Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "nll-landscape.png",
      "notebook_html": "logistic_nll_interactive.html",
      "notebook_ipynb": "logistic_nll_interactive.ipynb",
      "section_title": "Interactive Notebook: Logistic Models and NLL Landscape",
      "description": "Tune the logistic model parameters and dataset settings, then compare the current model against the minimum-NLL solution.",
      "iframe_title": "Interactive Logistic Regression and NLL Landscape Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "sign-reg.png",
      "notebook_html": "sign_classification_interactive.html",
      "notebook_ipynb": "sign_classification_interactive.ipynb",
      "section_title": "Interactive Notebook: Sign Function for Classification",
      "description": "Explore the sign-based binary classifier with adjustable class centers, spread, and decision boundary.",
      "iframe_title": "Interactive Sign Function Classification Notebook",
      "iframe_height": 860
    },
    {
      "image_match": "linearhyperplane.png",
      "notebook_html": "linear_hyperplane_interactive.html",
      "notebook_ipynb": "linear_hyperplane_interactive.ipynb",
      "section_title": "Interactive Notebook: Linear Separating Hyperplane",
      "description": "Adjust the weight vector and bias to see how the hyperplane changes in two dimensions.",
      "iframe_title": "Interactive Linear Separating Hyperplane Notebook",
      "iframe_height": 780
    },
    {
      "image_match": "img/lecture3/hinge-loss.png",
      "notebook_html": "hinge_loss_interactive.html",
      "notebook_ipynb": "hinge_loss_interactive.ipynb",
      "section_title": "Interactive Notebook: Hinge Loss Family",
      "description": "Covers hinge loss, hinge vs softplus, and subgradient with an adjustable bias term b.",
      "iframe_title": "Interactive Hinge Loss Family Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "img/lecture3/softplus.png",
      "notebook_html": "hinge_loss_interactive.html",
      "notebook_ipynb": "hinge_loss_interactive.ipynb",
      "section_title": "Interactive Notebook: Hinge Loss Family",
      "description": "Covers hinge loss, hinge vs softplus, and subgradient with an adjustable bias term b.",
      "iframe_title": "Interactive Hinge Loss Family Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "img/lecture3/subgrad.png",
      "notebook_html": "hinge_loss_interactive.html",
      "notebook_ipynb": "hinge_loss_interactive.ipynb",
      "section_title": "Interactive Notebook: Hinge Loss Family",
      "description": "Covers hinge loss, hinge vs softplus, and subgradient with an adjustable bias term b.",
      "iframe_title": "Interactive Hinge Loss Family Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "qa1-soft-hard.png",
      "notebook_html": "Lecture4_Q&A-Q1_interactive.html",
      "notebook_ipynb": "Lecture4_Q&A-Q1.ipynb",
      "section_title": "Interactive Notebook: Lecture 4 Q&A 1 Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 4 Q&A 1 Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "qa1-outlier-scatter.png",
      "notebook_html": "Lecture4_Q&A-Q1_interactive.html",
      "notebook_ipynb": "Lecture4_Q&A-Q1.ipynb",
      "section_title": "Interactive Notebook: Lecture 4 Q&A 1 Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 4 Q&A 1 Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "qa1-svms-outlier.png",
      "notebook_html": "Lecture4_Q&A-Q1_interactive.html",
      "notebook_ipynb": "Lecture4_Q&A-Q1.ipynb",
      "section_title": "Interactive Notebook: Lecture 4 Q&A 1 Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 4 Q&A 1 Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "qa2-nonlinear-2d.png",
      "notebook_html": "Lecture4_Q&A-Q2_interactive.html",
      "notebook_ipynb": "Lecture4_Q&A-Q2.ipynb",
      "section_title": "Interactive Notebook: Lecture 4 Q&A 2 Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 4 Q&A 2 Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "qa2-nonlinear-3d.png",
      "notebook_html": "Lecture4_Q&A-Q2_interactive.html",
      "notebook_ipynb": "Lecture4_Q&A-Q2.ipynb",
      "section_title": "Interactive Notebook: Lecture 4 Q&A 2 Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 4 Q&A 2 Notebook",
      "iframe_height": 820
    },
    {
      "image_match": "slpex-data.png",
      "notebook_html": "Lecture5_SLP-Example_interactive.html",
      "notebook_ipynb": "Lecture5_SLP-Example.ipynb",
      "section_title": "Interactive Notebook: Lecture 5: SLP Example Notebook",
      "description": "Run and edit this notebook interactively in JupyterLite directly from this lecture section.",
      "iframe_title": "Interactive Notebook - Lecture 5: SLP Example Notebook",
      "iframe_height": 800
    },
    {
      "image_match": "img/lecture6/threshold_tradeoff.png",
      "notebook_html": "Lecture6_ThresholdEffect_interactive.html",
      "notebook_ipynb": "Lecture6_ThresholdEffect.ipynb",
      "section_title": "Interactive Notebook: Figure 6.5 Precision and Recall vs Decision Threshold",
      "description": "Adjust the decision threshold and observe precision-recall trade-offs (Figure 6.5).",
      "iframe_title": "Interactive Notebook - Figure 6.5 Precision and Recall vs Threshold",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/nonlinear data generated via quadratic equation and polynomial regression model prediction fitted to data.png",
      "notebook_html": "Lecture8_Fig82_interactive.html",
      "notebook_ipynb": "Lecture8_Fig82_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 8.2 Polynomial Fit",
      "description": "Interactivity from linear to polynomial fit only (contour and optimization-MSE panels removed).",
      "iframe_title": "Interactive Figure 8.2 Polynomial Fit Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/contour-plot.png",
      "notebook_html": "Lecture8_LossContour_Interactive.html",
      "notebook_ipynb": "Lecture8_LossContour_Interactive.ipynb",
      "section_title": "Interactive Notebook: Loss Landscape and Contour Explorer",
      "description": "Pick a horizontal plane and observe how the corresponding contour level changes.",
      "iframe_title": "Interactive Loss Landscape and Contour Explorer",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/loss-landscape.png",
      "notebook_html": "Lecture8_LossContour_Interactive.html",
      "notebook_ipynb": "Lecture8_LossContour_Interactive.ipynb",
      "section_title": "Interactive Notebook: Loss Landscape and Contour Explorer",
      "description": "Pick a horizontal plane and observe how the corresponding contour level changes.",
      "iframe_title": "Interactive Loss Landscape and Contour Explorer",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/convergence path of batch gd.png",
      "notebook_html": "Lecture8_BatchGD_Interactive_interactive.html",
      "notebook_ipynb": "Lecture8_BatchGD_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 8.4 Batch GD",
      "description": "Explore convergence path with controllable alpha and stopping criteria.",
      "iframe_title": "Interactive Figure 8.4 Batch GD Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/convergence path of stochastic gd.png",
      "notebook_html": "Lecture8_SGD_Interactive_interactive.html",
      "notebook_ipynb": "Lecture8_SGD_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 8.5 Stochastic GD",
      "description": "Explore convergence path with controllable alpha and stopping criteria.",
      "iframe_title": "Interactive Figure 8.5 Stochastic GD Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/convergence path of mini-batch gd.png",
      "notebook_html": "Lecture8_MiniBatchGD_Interactive_interactive.html",
      "notebook_ipynb": "Lecture8_MiniBatchGD_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 8.6 Mini-batch GD",
      "description": "Explore convergence path with controllable alpha and stopping criteria.",
      "iframe_title": "Interactive Figure 8.6 Mini-batch GD Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture8/qa-mse-contour.png",
      "notebook_html": "Lecture8_Q4_Interactive_interactive.html",
      "notebook_ipynb": "Lecture8_Q4_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 8 Q4 — L1-Regularized MSE Contour",
      "description": "Drag λ to see how the L1 penalty deforms the elliptical MSE contours into axis-aligned kink shapes.",
      "iframe_title": "Interactive Lecture 8 Q4 Notebook",
      "iframe_height": 620
    },
    {
      "image_match": "img/lecture8/qa_q5_linear_vs_poly_fit.png",
      "notebook_html": "Lecture8_Q5_Interactive_interactive.html",
      "notebook_ipynb": "Lecture8_Q5_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 8 Q5 — Linear vs Polynomial Regression",
      "description": "Adjust noise and polynomial degree to compare linear and polynomial regression fits on sinusoidal data.",
      "iframe_title": "Interactive Lecture 8 Q5 Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture9/p9.png",
      "notebook_html": "Lecture9_Fig98_OLS_Noise_interactive.html",
      "notebook_ipynb": "Lecture9_Fig98_OLS_Noise.ipynb",
      "section_title": "Interactive Notebook: Figure 9.8 Noise Sensitivity (OLS)",
      "description": "Control noise and compare coefficient variability under OLS.",
      "iframe_title": "Interactive Figure 9.8 Noise Sensitivity (OLS) Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture9/p10.png",
      "notebook_html": "Lecture9_Fig99_Ridge_Noise_interactive.html",
      "notebook_ipynb": "Lecture9_Fig99_Ridge_Noise.ipynb",
      "section_title": "Interactive Notebook: Figure 9.9 Ridge Under Noise",
      "description": "Control noise and compare ridge robustness to OLS.",
      "iframe_title": "Interactive Figure 9.9 Ridge Under Noise Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture9/p11.png",
      "notebook_html": "Lecture9_Fig910_SampleCount_interactive.html",
      "notebook_ipynb": "Lecture9_Fig910_SampleCount.ipynb",
      "section_title": "Interactive Notebook: Figure 9.10 Sample Count Effect",
      "description": "Control number of data points and inspect fit stability.",
      "iframe_title": "Interactive Figure 9.10 Sample Count Effect Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture9/p12.png",
      "notebook_html": "Lecture9_Fig911_LambdaEffect_interactive.html",
      "notebook_ipynb": "Lecture9_Fig911_LambdaEffect.ipynb",
      "section_title": "Interactive Notebook: Figure 9.11 Lambda Effect",
      "description": "Adjust λ and observe coefficient shrinkage behavior.",
      "iframe_title": "Interactive Figure 9.11 Lambda Effect Notebook",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture9/p13.png",
      "notebook_html": "Lecture9_Fig912_LambdaBehavior_interactive.html",
      "notebook_ipynb": "Lecture9_Fig912_LambdaBehavior.ipynb",
      "section_title": "Interactive Notebook: Figure 9.12 Lambda vs Model Behavior",
      "description": "Adjust λ and inspect resulting model complexity.",
      "iframe_title": "Interactive Figure 9.12 Lambda vs Model Behavior Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture10/running_example.png",
      "notebook_html": "Lecture10_Mahalanobis_Distance_Visualization_interactive.html",
      "notebook_ipynb": "Lecture10_Mahalanobis_Distance_Visualization.ipynb",
      "section_title": "Interactive Notebook: Table 10.1 Data-Point Editor",
      "description": "Select a point, move it with sliders, and switch between the Table 10.2 to 10.5 metric views one at a time.",
      "iframe_title": "Interactive Table 10.1 Data-Point Editor Notebook",
      "iframe_height": 920,
      "replace_start_id": "running-example-used-throughout-this-section"
    },
    {
      "image_match": "img/lecture10/case5_rotated.png",
      "notebook_html": "Lecture10_Sigma_Interactive_interactive.html",
      "notebook_ipynb": "Lecture10_Sigma_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 10.9 Sigma Control",
      "description": "Adjust covariance matrix Sigma after Case 5 and observe Mahalanobis geometry changes.",
      "iframe_title": "Interactive Figure 10.9 Sigma Control Notebook",
      "iframe_height": 1060
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 10.43.43 am.png",
      "notebook_html": "Lecture11_KMeans_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMeans_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 11.1 K-Means Steps",
      "description": "Choose initial centroids directly on the plot and walk through assignment and update steps.",
      "iframe_title": "Interactive Figure 11.1 K-Means Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 11.12.16 am.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 11.14.26 am.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 11.31.47 am.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 12.07.38 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 12.10.20 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 12.18.27 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 1.38.32 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 1.41.59 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 2.01.26 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 2.21.34 pm.png",
      "notebook_html": "Lecture11_KMeans_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMeans_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 11.11 Choosing k",
      "description": "Vary the number of clusters k and see how the assignment pattern changes.",
      "iframe_title": "Interactive Figure 11.11 Choosing k Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/screenshot 2024-09-29 at 2.23.47 pm.png",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/images/dataset.jpg",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/images/undesirable_clustering.jpg",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/images/ideal_clustering.jpg",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/images/mini_batch_comp.jpg",
      "notebook_html": "Lecture11_MiniBatch_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_MiniBatch_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figure 11.14 Mini-batch Optimization",
      "description": "Compare standard and mini-batch k-means under the same initialization and inspect the optimization tradeoff.",
      "iframe_title": "Interactive Figure 11.14 Mini-batch Optimization Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/mini-batch-k-means-diff.jpg",
      "notebook_html": "",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Lecture 11 K-Means Steps",
      "description": "Choose initial centroids, run k-means one step at a time, and vary k.",
      "iframe_title": "Interactive K-Means Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture11/images/k_medoids_data.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/k-med-init-assign.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_1_a_table.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_1_a_split.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/k-med-update-1-b-table.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_1_b_split.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/k-med-reassign.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_2_a_table.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/udpate_2_a_graph.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_2_b_table.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture11/images/update_2_b_graph.jpg",
      "notebook_html": "Lecture11_KMedoids_Interactive_interactive.html",
      "notebook_ipynb": "Lecture11_KMedoids_Interactive.ipynb",
      "section_title": "Interactive Notebook: Figures 11.16 to 11.26 k-Medoids Walkthrough",
      "description": "Step through the Manhattan-distance k-medoids example and inspect candidate medoid costs at each stage.",
      "iframe_title": "Interactive Figures 11.16 to 11.26 k-Medoids Notebook",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture12/gmm clustering examples/initialization of em algorithm.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/gmm clustering examples/iteration 0 of em algorithm.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/gmm clustering examples/iteration 50 of em algorithm.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/gmm clustering examples/iteration 100 of em algorithm.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/gmm clustering examples/iteration 150 of em algorithm.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/clustering performance evaluation/silhouette coefficient.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/clustering performance evaluation/cluster silhouette plot.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture12/clustering performance evaluation/rand index.png",
      "notebook_html": "Lecture12_Metrics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture12_Metrics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Lecture 12.3 Clustering Metrics",
      "description": "Compute internal and external clustering metrics on the same data and compare outcomes.",
      "iframe_title": "Interactive Clustering Metrics Notebook",
      "iframe_height": 680
    },
    {
      "image_match": "img/lecture7/mse_contour_theta0_theta1.png",
      "notebook_html": "Contour_interactive.html",
      "notebook_ipynb": "Contour.ipynb",
      "section_title": "Interactive Notebook: Figure 7.3 Least Squares Loss Contours",
      "description": "Explore how changing theta0 and theta1 moves the solution on the least-squares contour surface.",
      "iframe_title": "Interactive Notebook - Figure 7.3 Loss Contour",
      "iframe_height": 700
    },
    {
      "image_match": "img/lecture7/image.png",
      "notebook_html": "Lecture7_Figure74_interactive.html",
      "notebook_ipynb": "Lecture7_Figure74.ipynb",
      "section_title": "Interactive Notebook: Figure 7.4 Linear Subspace Geometry",
      "description": "Interact with the 2D subspace spanned by basis vectors and inspect geometric relationships.",
      "iframe_title": "Interactive Notebook - Figure 7.4 Linear Subspace Geometry",
      "iframe_height": 700
    },
    {
      "image_match": "img/lecture7/projectionmatrix.png",
      "notebook_html": "Lecture7_Figure75_interactive.html",
      "notebook_ipynb": "Lecture7_Figure75.ipynb",
      "section_title": "Interactive Notebook: Projection and Residual Orthogonality",
      "description": "Adjust vectors and observe projection of y onto the subspace with orthogonal residual y-hat.",
      "iframe_title": "Interactive Notebook - Figure 7.5 Projection and Residual Orthogonality",
      "iframe_height": 700
    },
    {
      "image_match": "img/lecture6/cur.png",
      "notebook_html": "Lecture6_ThresholdEffect_interactive.html",
      "notebook_ipynb": "Lecture6_ThresholdEffect.ipynb",
      "section_title": "Interactive Notebook: Figure 6.5 Precision and Recall vs Decision Threshold",
      "description": "Adjust the decision threshold and observe precision-recall trade-offs (Figure 6.5).",
      "iframe_title": "Interactive Notebook - Figure 6.5 Precision and Recall vs Threshold",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture9/p15.png",
      "notebook_html": "Lecture9_Fig915_OptimizationPath_interactive.html",
      "notebook_ipynb": "Lecture9_Fig915_OptimizationPath.ipynb",
      "section_title": "Interactive Notebook: Figure 9.15 Geometric Intuition for L1, L2, and Elastic Net",
      "description": "Explore the L1, L2, and Elastic Net constraint geometry together with their optimization paths.",
      "iframe_title": "Interactive Figure 9.15 Geometric Intuition for L1, L2, and Elastic Net Notebook",
      "iframe_height": 980
    },
    {
      "image_match": "img/lecture9/screenshot*6.22.40*",
      "notebook_html": "Lecture9_Fig920_ModelComplexity_interactive.html",
      "notebook_ipynb": "Lecture9_Fig920_ModelComplexity.ipynb",
      "section_title": "Interactive Notebook: Figure 9.20 Model Complexity",
      "description": "Adjust model complexity and observe error behavior.",
      "iframe_title": "Interactive Figure 9.20 Model Complexity Notebook",
      "iframe_height": 1080
    },
    {
      "image_match": "img/lecture9/screenshot*6.28.08*",
      "notebook_html": "Lecture9_Fig921_ComplexityVsError_interactive.html",
      "notebook_ipynb": "Lecture9_Fig921_ComplexityVsError.ipynb",
      "section_title": "Interactive Notebook: Figure 9.21 Model Complexity vs Error",
      "description": "Adjust model complexity and inspect bias/variance trends.",
      "iframe_title": "Interactive Figure 9.21 Model Complexity vs Error Notebook",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture9/p16.png",
      "notebook_html": "Lecture9_Fig925_GTNoise_interactive.html",
      "notebook_ipynb": "Lecture9_Fig925_GTNoise.ipynb",
      "section_title": "Interactive Notebook: Figure 9.25 Measured vs True Accuracy Under Label Noise",
      "description": "Adjust label noise and observed accuracy to inspect the range of plausible true accuracies.",
      "iframe_title": "Interactive Figure 9.25 Measured vs True Accuracy Under Label Noise Notebook",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture14/image4.png",
      "notebook_html": "Lecture13_Kernel_Interactive_interactive.html",
      "notebook_ipynb": "Lecture13_Kernel_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 4.1 — 2D Convolution Kernel Demo",
      "description": "Adjust the kernel size and values, then step through positions to see how each output value is computed from the element-wise product.",
      "iframe_title": "Interactive Section 4.1 Convolution Kernel Demo",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture14/12.png",
      "notebook_html": "Lecture13_OutputSize_Interactive_interactive.html",
      "notebook_ipynb": "Lecture13_OutputSize_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.2 — Output Size Formula",
      "description": "Vary input size N, filter size F, stride S, and padding P to see the output dimensions change live. Switch to non-square mode for separate height and width controls.",
      "iframe_title": "Interactive Section 6.2 Output Size Formula",
      "iframe_height": 820
    },
    {
      "image_match": "img/lecture14/13.png",
      "notebook_html": "Lecture13_Stride_Interactive_interactive.html",
      "notebook_ipynb": "Lecture13_Stride_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.3 — Impact of Stride",
      "description": "Change the stride to see which positions the filter visits and how skipping positions reduces the output size.",
      "iframe_title": "Interactive Section 6.3 Impact of Stride",
      "iframe_height": 720
    },
    {
      "image_match": "img/lecture14/16.png",
      "notebook_html": "Lecture13_Padding_Interactive_interactive.html",
      "notebook_ipynb": "Lecture13_Padding_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.4 — Zero-Padding",
      "description": "Adjust zero-padding P to see the padded input and resulting output size. Observe the special case where padding achieves \"same\" convolution.",
      "iframe_title": "Interactive Section 6.4 Zero-Padding",
      "iframe_height": 780
    },
    {
      "image_match": "img/lecture14/Im2.png",
      "notebook_html": "Lecture13_Pooling_Interactive_interactive.html",
      "notebook_ipynb": "Lecture13_Pooling_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 7 — Pooling Layer",
      "description": "Choose between max, average, and L2 pooling. Step through the pooling window to see how each operation downsamples the feature map.",
      "iframe_title": "Interactive Section 7 Pooling Layer",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture15/vanishing gradient problem.png",
      "notebook_html": "Lecture14_VanishingGradient_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_VanishingGradient_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 5 — Vanishing Gradient Problem",
      "description": "See how gradient magnitudes shrink through deep networks for different activation functions. Toggle activations and adjust depth to observe vanishing vs stable gradients.",
      "iframe_title": "Interactive Section 5 Vanishing Gradient Problem",
      "iframe_height": 820
    },
    {
      "image_match": "img/lecture15/AlexNet3_2.png",
      "notebook_html": "Lecture14_Dropout_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_Dropout_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.3 — Dropout Regularization",
      "description": "Click any hidden neuron to toggle it on/off and simulate dropout. Track the effective dropout rate and see how the active sub-network changes.",
      "iframe_title": "Interactive Section 6.3 Dropout Regularization",
      "iframe_height": 720
    },
    {
      "image_match": "img/lecture15/AlexNet4_1.png",
      "notebook_html": "Lecture14_TrainingDynamics_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_TrainingDynamics_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.4.2 — Training Dynamics",
      "description": "Train a small neural network on a toy dataset and compare how different activation functions affect convergence speed and final accuracy.",
      "iframe_title": "Interactive Section 6.4.2 Training Dynamics",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture15/AlexNet6_2.png",
      "notebook_html": "Lecture14_InterChannelNorm_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_InterChannelNorm_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6.5.4 — Inter-Channel Normalization (LRN)",
      "description": "Adjust LRN parameters k, α, β, and window n to see how channel activations are normalized. Observe how large activations are suppressed more than small ones.",
      "iframe_title": "Interactive Section 6.5.4 Inter-Channel LRN",
      "iframe_height": 740
    },
    {
      "image_match": "img/lecture15/BatchNormalization.png",
      "notebook_html": "Lecture14_BatchNorm_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_BatchNorm_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 8.3.0.1 — Batch Normalizing Transform",
      "description": "Adjust batch size, γ (scale), β (shift), and ε to see the three-step BN transform: compute statistics, normalize, then scale and shift.",
      "iframe_title": "Interactive Section 8.3.0.1 Batch Normalizing Transform",
      "iframe_height": 780
    },
    {
      "image_match": "img/lecture15/ResBlock.png",
      "notebook_html": "Lecture14_ResidualConnection_Interactive_interactive.html",
      "notebook_ipynb": "Lecture14_ResidualConnection_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 9.1 — Residual Connections",
      "description": "Compare gradient flow in plain vs residual networks. Vary depth and weight scale to see how skip connections prevent vanishing gradients.",
      "iframe_title": "Interactive Section 9.1 Residual Connections",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture16/saddle_point.jpg",
      "notebook_html": "Lecture16_SaddlePoint_Interactive_interactive.html",
      "notebook_ipynb": "Lecture16_SaddlePoint_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 4.2 — Saddle Points",
      "description": "Gradient descent locks onto the saddle point of f(x,y)=x²−y² regardless of learning rate. Toggle modes to see how noise or momentum can escape.",
      "iframe_title": "Interactive Section 4.2 Saddle Points",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture16/gradient_plus_mumentum.jpg",
      "notebook_html": "Lecture16_OptimizerComparison_Interactive_interactive.html",
      "notebook_ipynb": "Lecture16_OptimizerComparison_Interactive.ipynb",
      "section_title": "Interactive Notebook: Sections 5–8 — Optimizer Comparison",
      "description": "Compare Batch GD, Momentum, AdaGrad, and RMSProp on ill-conditioned loss landscapes. Toggle each optimizer and tune parameters independently.",
      "iframe_title": "Interactive Sections 5–8 Optimizer Comparison",
      "iframe_height": 1020
    },
    {
      "image_match": "img/lecture16/Adagrad.jpg",
      "notebook_html": "Lecture16_OptimizerComparison_Interactive_interactive.html",
      "notebook_ipynb": "Lecture16_OptimizerComparison_Interactive.ipynb",
      "section_title": "Interactive Notebook: Sections 5–8 — Optimizer Comparison",
      "description": "Compare Batch GD, Momentum, AdaGrad, and RMSProp on ill-conditioned loss landscapes. Toggle each optimizer and tune parameters independently.",
      "iframe_title": "Interactive Sections 5–8 Optimizer Comparison",
      "iframe_height": 1020
    },
    {
      "image_match": "img/lecture16/Curvature.png",
      "notebook_html": "Lecture16_ConditionNumber_Interactive_interactive.html",
      "notebook_ipynb": "Lecture16_ConditionNumber_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 6 — Condition Numbers & Curvature",
      "description": "Adjust the Hessian eigenvalues λ₁ and λ₂ to change the condition number κ=λ_max/λ_min. See how anisotropic curvature slows GD convergence.",
      "iframe_title": "Interactive Section 6 Condition Numbers",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture16/RMSProp.png",
      "notebook_html": "Lecture16_Adam_Interactive_interactive.html",
      "notebook_ipynb": "Lecture16_Adam_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 9 — Adam Optimizer",
      "description": "Adam combines momentum and RMSProp with bias correction. Adjust α, β₁, β₂ and compare against GD, Momentum, and RMSProp.",
      "iframe_title": "Interactive Section 9 Adam Optimizer",
      "iframe_height": 960
    },
    {
      "image_match": "img/lecture17/scribeim3.png",
      "notebook_html": "Lecture17_LossFunctions_Interactive_interactive.html",
      "notebook_ipynb": "Lecture17_LossFunctions_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 4 — CE vs Weighted CE vs Focal Loss",
      "description": "Adjust class weights (WCE) and focus parameter γ (Focal Loss) to see how each loss re-weights hard vs easy examples for imbalanced datasets.",
      "iframe_height": 840,
      "iframe_title": "Interactive Section 4 Loss Functions"
    },
    {
      "image_match": "img/lecture17/scribeim7.png",
      "notebook_html": "Lecture17_TrainingPlayground_Interactive_interactive.html",
      "notebook_ipynb": "Lecture17_TrainingPlayground_Interactive.ipynb",
      "section_title": "Interactive Notebook: Sections 10–11 — Weight Init & Activation/Gradient Histograms",
      "description": "Select initialization scheme (Zeros, Xavier, Kaiming) and activation function to see activation std and gradient norms per layer. Compare healthy vs broken configurations.",
      "iframe_height": 980,
      "iframe_title": "Interactive Sections 10-11 Weight Init and Activation Histograms"
    },
    {
      "image_match": "img/lecture18/output.png",
      "notebook_html": "Lecture18_PCA_Interactive_interactive.html",
      "notebook_ipynb": "Lecture18_PCA_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 2.2 — PCA",
      "description": "Explore principal components: adjust data angle, variance ratio, and number of points. See how PC arrows align with maximum variance directions and how a linear AE learns the same subspace.",
      "iframe_height": 820,
      "iframe_title": "Interactive Section 2.2 PCA"
    },
    {
      "image_match": "img/lecture18/screenshot 2024-10-30 at 6.37.17 pm.png",
      "notebook_html": "Lecture18_AE_Interactive_interactive.html",
      "notebook_ipynb": "Lecture18_AE_Interactive.ipynb",
      "section_title": "Interactive Notebook: Sections 2.3–3 — Linear vs Nonlinear Autoencoder",
      "description": "Compare linear AE (equivalent to PCA) with nonlinear AEs (Sigmoid, Tanh, ReLU). Adjust latent dimension k and see reconstruction quality and latent space cluster separation.",
      "iframe_height": 920,
      "iframe_title": "Interactive Sections 2.3-3 Linear vs Nonlinear AE"
    },
    {
      "image_match": "img/lecture18/screenshot 2024-11-02 at 9.41.54 pm.png",
      "notebook_html": "Lecture18_Unpooling_Interactive_interactive.html",
      "notebook_ipynb": "Lecture18_Unpooling_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 5.1 — Unpooling Strategies",
      "description": "See all three unpooling strategies side by side: max-unpooling (uses stored indices), bed-of-nails (top-left placement), and nearest-neighbor (fill entire block). Randomize the input to explore different cases.",
      "iframe_height": 860,
      "iframe_title": "Interactive Section 5.1 Unpooling"
    },
    {
      "image_match": "img/lecture18/screenshot 2024-11-02 at 8.32.21 pm.png",
      "notebook_html": "Lecture18_TransposedConv_Interactive_interactive.html",
      "notebook_ipynb": "Lecture18_TransposedConv_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 5.3 — Transposed Convolution",
      "description": "Step through a transposed convolution one input element at a time. See how each input scales the kernel and accumulates in the output. Formula: H_out = (H_in−1)×s − 2p + k.",
      "iframe_height": 960,
      "iframe_title": "Interactive Section 5.3 Transposed Convolution"
    },
    {
      "image_match": "img/lecture18/screenshot 2024-11-02 at 8.28.20 pm.png",
      "notebook_html": "Lecture18_ConvAE_Interactive_interactive.html",
      "notebook_ipynb": "Lecture18_ConvAE_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 5.4 — Convolutional Autoencoder on MNIST",
      "description": "Compare FC-AE and Conv-AE reconstruction quality across latent dimensions. Conv-AEs exploit spatial structure for better reconstructions at smaller latent sizes.",
      "iframe_height": 980,
      "iframe_title": "Interactive Section 5.4 Convolutional AE"
    },
    {
      "image_match": "img/lecture19/trick2.jpg",
      "notebook_html": "Lecture19_VAETraining_Interactive_interactive.html",
      "notebook_ipynb": "Lecture19_VAETraining_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 2.4 — VAE Training (ELBO & Reparameterization)",
      "description": "Adjust μ and σ to see how the posterior q(z|x) compares to the prior N(0,1). The KL divergence measures their distance. See the reparameterization trick z=μ+σε and the full ELBO as a function of μ.",
      "iframe_height": 900,
      "iframe_title": "Interactive Section 2.4 VAE Training"
    },
    {
      "image_match": "img/lecture19/novelsamples.jpg",
      "notebook_html": "Lecture19_VAEInference_Interactive_interactive.html",
      "notebook_ipynb": "Lecture19_VAEInference_Interactive.ipynb",
      "section_title": "Interactive Notebook: Section 2.4 — VAE Latent Space Explorer",
      "description": "Click anywhere in the 2D latent space to decode z directly into a digit — no encoder needed. Explore how different regions generate different digits and interpolate smoothly between them.",
      "iframe_height": 860,
      "iframe_title": "Interactive Section 2.4 VAE Inference"
    },
    {
      "image_match": "img/lecture20/Gradient.png",
      "notebook_html": "Lecture20_GradientVanishing_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3.3 — Vanishing Gradient in RNNs",
      "description": "Vary the number of timesteps T and weight magnitude to see how gradient magnitude decays for sigmoid, tanh, and ReLU activations in BPTT.",
      "iframe_title": "Interactive Section 3.3 Vanishing Gradient in RNNs",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture20/ReLU.png",
      "notebook_html": "Lecture20_GradientVanishing_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3.3 — Vanishing Gradient in RNNs",
      "description": "Vary the number of timesteps T and weight magnitude to see how gradient magnitude decays for sigmoid, tanh, and ReLU activations in BPTT.",
      "iframe_title": "Interactive Section 3.3 Vanishing Gradient in RNNs",
      "iframe_height": 760
    },
    {
      "image_match": "img/lecture20/LSTM.png",
      "notebook_html": "Lecture20_LSTMGates_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 4–5 — LSTM and GRU Gate Mechanics",
      "description": "Toggle forget, input, and output gates on/off and adjust inputs to see how cell state and hidden state evolve over time. Switch to GRU tab for update/reset gates.",
      "iframe_title": "Interactive Sections 4–5 LSTM and GRU Gates",
      "iframe_height": 780
    },
    {
      "image_match": "img/lecture20/global_attention.png",
      "notebook_html": "Lecture21_Attention_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Local vs Global Attention",
      "description": "Click any token as the query and compare global (all-positions) vs local (window D) attention scores. See the full NxN attention heatmap for both modes.",
      "iframe_title": "Interactive Section 3 Local vs Global Attention",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture20/local_attention.png",
      "notebook_html": "Lecture21_Attention_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Local vs Global Attention",
      "description": "Click any token as the query and compare global (all-positions) vs local (window D) attention scores. See the full NxN attention heatmap for both modes.",
      "iframe_title": "Interactive Section 3 Local vs Global Attention",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture26/SimCLR_Framework.png",
      "notebook_html": "Lecture25_SSL_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Self-Supervised Learning & SimCLR",
      "description": "Explore pretext tasks (rotation, masking, jigsaw, colorization) and SimCLR contrastive learning. See how the latent space clusters improve with more SSL pretraining.",
      "iframe_title": "Interactive Section 3 SSL and SimCLR",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture26/cont.png",
      "notebook_html": "Lecture25_SSL_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Self-Supervised Learning & SimCLR",
      "description": "Explore pretext tasks (rotation, masking, jigsaw, colorization) and SimCLR contrastive learning. See how the latent space clusters improve with more SSL pretraining.",
      "iframe_title": "Interactive Section 3 SSL and SimCLR",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture25/entro_sampling.png",
      "notebook_html": "Lecture26_ActiveLearning_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 6.2 — Active Learning Sampling Strategies",
      "description": "Compare entropy, margin, and least-confidence sampling on a 2D dataset. Query the most uncertain points batch by batch and watch the labeled set grow.",
      "iframe_title": "Interactive Section 6.2 Active Learning Sampling",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture25/margin_sampling.png",
      "notebook_html": "Lecture26_ActiveLearning_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 6.2 — Active Learning Sampling Strategies",
      "description": "Compare entropy, margin, and least-confidence sampling on a 2D dataset. Query the most uncertain points batch by batch and watch the labeled set grow.",
      "iframe_title": "Interactive Section 6.2 Active Learning Sampling",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture25/least_sampling.png",
      "notebook_html": "Lecture26_ActiveLearning_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 6.2 — Active Learning Sampling Strategies",
      "description": "Compare entropy, margin, and least-confidence sampling on a 2D dataset. Query the most uncertain points batch by batch and watch the labeled set grow.",
      "iframe_title": "Interactive Section 6.2 Active Learning Sampling",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture27/factors_uncer.png",
      "notebook_html": "Lecture27_Uncertainty_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Aleatoric, Epistemic & Predictive Uncertainty",
      "description": "Visualize aleatoric (irreducible noise), epistemic (model/data), and predictive uncertainty on 1D regression. Explore MC Dropout uncertainty decomposition.",
      "iframe_title": "Interactive Section 3 Uncertainty Types",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture27/MC-dropout.jpg",
      "notebook_html": "Lecture27_Uncertainty_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Aleatoric, Epistemic & Predictive Uncertainty",
      "description": "Visualize aleatoric (irreducible noise), epistemic (model/data), and predictive uncertainty on 1D regression. Explore MC Dropout uncertainty decomposition.",
      "iframe_title": "Interactive Section 3 Uncertainty Types",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture27/uncertainty-summary.jpg",
      "notebook_html": "Lecture27_Uncertainty_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 3 — Aleatoric, Epistemic & Predictive Uncertainty",
      "description": "Visualize aleatoric (irreducible noise), epistemic (model/data), and predictive uncertainty on 1D regression. Explore MC Dropout uncertainty decomposition.",
      "iframe_title": "Interactive Section 3 Uncertainty Types",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture22/salient.png",
      "notebook_html": "Lecture23_Saliency_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 4 — Saliency and Attribution Methods",
      "description": "Compare occlusion-based, vanilla gradient, guided backpropagation, SmoothGrad, and integrated gradients saliency maps on synthetic input images.",
      "iframe_title": "Interactive Section 4 Saliency Methods",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture22/forward.png",
      "notebook_html": "Lecture23_Saliency_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 4 — Saliency and Attribution Methods",
      "description": "Compare occlusion-based, vanilla gradient, guided backpropagation, SmoothGrad, and integrated gradients saliency maps on synthetic input images.",
      "iframe_title": "Interactive Section 4 Saliency Methods",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture22/xdog.png",
      "notebook_html": "Lecture23_Saliency_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 4 — Saliency and Attribution Methods",
      "description": "Compare occlusion-based, vanilla gradient, guided backpropagation, SmoothGrad, and integrated gradients saliency maps on synthetic input images.",
      "iframe_title": "Interactive Section 4 Saliency Methods",
      "iframe_height": 860
    },
    {
      "image_match": "img/lecture22/activation.png",
      "notebook_html": "Lecture23_GradCAM_LIME_SHAP_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 5–6 — GradCAM, LIME, and SHAP",
      "description": "Explore class activation maps (GradCAM), perturbation-based superpixel explanations (LIME), and Shapley value attributions (SHAP) on synthetic images.",
      "iframe_title": "Interactive Sections 5–6 GradCAM LIME SHAP",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture22/lime_pipeline.png",
      "notebook_html": "Lecture23_GradCAM_LIME_SHAP_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 5–6 — GradCAM, LIME, and SHAP",
      "description": "Explore class activation maps (GradCAM), perturbation-based superpixel explanations (LIME), and Shapley value attributions (SHAP) on synthetic images.",
      "iframe_title": "Interactive Sections 5–6 GradCAM LIME SHAP",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture22/shap.png",
      "notebook_html": "Lecture23_GradCAM_LIME_SHAP_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 5–6 — GradCAM, LIME, and SHAP",
      "description": "Explore class activation maps (GradCAM), perturbation-based superpixel explanations (LIME), and Shapley value attributions (SHAP) on synthetic images.",
      "iframe_title": "Interactive Sections 5–6 GradCAM LIME SHAP",
      "iframe_height": 900
    },
    {
      "image_match": "img/lecture23/net2.png",
      "notebook_html": "Lecture24_ContrastiveCAM_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 2–4 — Contrastive CAM, Counterfactual CAM & XAI Evaluation",
      "description": "Ask 'Why P rather than Q?' with Contrastive-CAM, explore counterfactual image modifications, and compare XAI methods by faithfulness, sensitivity, and human agreement.",
      "iframe_title": "Interactive Sections 2–4 Contrastive CAM and XAI Evaluation",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture23/net10.png",
      "notebook_html": "Lecture24_ContrastiveCAM_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 2–4 — Contrastive CAM, Counterfactual CAM & XAI Evaluation",
      "description": "Ask 'Why P rather than Q?' with Contrastive-CAM, explore counterfactual image modifications, and compare XAI methods by faithfulness, sensitivity, and human agreement.",
      "iframe_title": "Interactive Sections 2–4 Contrastive CAM and XAI Evaluation",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture22/vsi.png",
      "notebook_html": "Lecture24_ContrastiveCAM_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 2–4 — Contrastive CAM, Counterfactual CAM & XAI Evaluation",
      "description": "Ask 'Why P rather than Q?' with Contrastive-CAM, explore counterfactual image modifications, and compare XAI methods by faithfulness, sensitivity, and human agreement.",
      "iframe_title": "Interactive Sections 2–4 Contrastive CAM and XAI Evaluation",
      "iframe_height": 920
    },
    {
      "image_match": "img/lecture24/Picture24.5.png",
      "notebook_html": "Lecture28_AnomalyDetection_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 2–3 — ROC Curve, Threshold Trade-off & Detection Methods",
      "description": "Explore how the decision threshold γ controls TPR vs FPR, visualize the ROC curve and AUC, and compare distance-based, Isolation Forest, and density-based anomaly detection on 2D data.",
      "iframe_title": "Interactive Anomaly Detection: ROC, Threshold and Detection Methods",
      "iframe_height": 880
    },
    {
      "image_match": "img/lecture24/Picture6.png",
      "notebook_html": "Lecture28_AnomalyDetection_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Sections 2–3 — ROC Curve, Threshold Trade-off & Detection Methods",
      "description": "Explore how the decision threshold γ controls TPR vs FPR, visualize the ROC curve and AUC, and compare distance-based, Isolation Forest, and density-based anomaly detection on 2D data.",
      "iframe_title": "Interactive Anomaly Detection: ROC, Threshold and Detection Methods",
      "iframe_height": 880
    },
    {
      "image_match": "img/lecture24/Picture11.png",
      "notebook_html": "Lecture28_AnomalyDetection_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Interactive Notebook: Section 4 — Reconstruction-Based Anomaly Detection",
      "description": "Simulate autoencoder reconstruction error for normal vs anomalous patches. Adjust latent dimension and error threshold to see how the bottleneck capacity affects detection sensitivity.",
      "iframe_title": "Interactive Autoencoder Reconstruction-based Anomaly Detection",
      "iframe_height": 880
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "regression",
      "notebook_html": "Lecture8_Polynomial_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — Polynomial Regression",
      "description": "Look ahead to the Polynomial Regression lecture: fit a polynomial of varying degree to noisy data and watch the bias–variance trade-off play out.",
      "iframe_title": "Preview: Polynomial Regression",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "classification",
      "notebook_html": "Lecture3_LinearBoundaries_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — Linear Decision Boundaries",
      "description": "Look ahead to Logistic Regression: drag the decision boundary to see how a linear classifier separates two classes.",
      "iframe_title": "Preview: Linear Decision Boundaries",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "clustering",
      "notebook_html": "Lecture11_KMeans_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — K-Means Clustering",
      "description": "Look ahead to K-Means: pick initial centroids and step through the assign/update iterations.",
      "iframe_title": "Preview: K-Means Clustering",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "dimensionality-reduction",
      "notebook_html": "Lecture18_PCA_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — PCA Dimensionality Reduction",
      "description": "Look ahead to PCA: project a high-dimensional toy dataset onto its top principal components.",
      "iframe_title": "Preview: PCA",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "core-linear-algebra-operations",
      "notebook_html": "Lecture16_ConditionNumber_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — Matrix Transformation (matrix-vector multiplication)",
      "description": "Apply a 2×2 matrix to the unit circle and see how matrix-vector multiplication stretches, rotates, and shears the input space.",
      "iframe_title": "Preview: Matrix Transformation",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "important-distributions",
      "notebook_html": "Lecture10_Sigma_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — Multivariate Gaussian (revisited under GMMs)",
      "description": "Adjust the covariance matrix Σ and watch the bivariate Gaussian iso-density ellipse change shape and orientation.",
      "iframe_title": "Preview: Multivariate Gaussian Σ",
      "iframe_height": 720
    },
    {
      "lecture_match": "Lecture2_Introduction-to-Machine-Learning",
      "section_id": "synthetic-datasets",
      "notebook_html": "Lecture14_TrainingDynamics_Interactive_interactive.html",
      "notebook_ipynb": "",
      "section_title": "Preview Demo — Synthetic Dataset Generation (CNN Architectures §6.4.2)",
      "description": "From the CNN Architectures lecture: see how a small synthetic dataset is generated and used to train a network. Focus on the data-generation portion at the top.",
      "iframe_title": "Preview: Synthetic Data Generation",
      "iframe_height": 760
    },
    {
      "lecture_match": "Lecture31_Large-Language-Models",
      "section_id": "context-length-in-modern-llms",
      "notebook_html": "Lecture28_SelfAttentionFlow_Interactive_interactive.html",
      "section_title": "Interactive Demo — Self-Attention Flow, Context Length & KV Caching",
      "description": "Walk through Q/K/V projections, attention weights, and the output. Tab 2 plots quadratic attention vs linear projection cost as context length grows; Tab 3 contrasts naive autoregressive generation with KV caching to show the O(T²) → O(T) saving. Inspired by ByHand AI's self-attention-flow visualization.",
      "iframe_title": "Self-Attention Flow & KV Caching",
      "iframe_height": 880
    },
    {
      "lecture_match": "Lecture31_Large-Language-Models",
      "section_id": "parameter-efficient-fine-tuning-with-lora",
      "notebook_html": "Lecture28_LoRA_Interactive_interactive.html",
      "section_title": "Interactive Demo — Low-Rank Adaptation (LoRA)",
      "description": "Vary rank r, scaling α, and matrix dimensions to see how the LoRA update ΔW = (α/r)·BA approximates a full fine-tuning delta. Tab 2 shows parameter savings vs full fine-tuning; Tab 3 plots reconstruction error of a rank-r approximation against the full target via truncated SVD.",
      "iframe_title": "LoRA Interactive",
      "iframe_height": 880
    },
    {
      "lecture_match": "Lecture31_Large-Language-Models",
      "section_id": "kv-caching-for-efficient-inference",
      "notebook_html": "Lecture28_SelfAttentionFlow_Interactive_interactive.html",
      "section_title": "Interactive Demo — KV Caching (Tab 3 of the Self-Attention Flow notebook)",
      "description": "The same Self-Attention Flow notebook used in §5.2 also illustrates KV caching. Open Tab 3 and step through new-token generation with caching on vs off; the FLOP counter shows the O(T²) → O(T) saving in real time.",
      "iframe_title": "KV Caching",
      "iframe_height": 880
    },
    {
      "lecture_match": "Lecture31_Large-Language-Models",
      "section_id": "illustrative-comparison-of-evaluation-metrics",
      "notebook_html": "Lecture28_EvalMetrics_Interactive_interactive.html",
      "section_title": "Interactive Demo — LLM Evaluation Metrics",
      "description": "Edit the reference and candidate text and watch BLEU-1..4, ROUGE-1/2/L, perplexity (from a tiny on-page bigram model), edit distance, and exact match update live with worked-example tables. Inspired by ByHand AI's AI-eval-equations explainer.",
      "iframe_title": "LLM Evaluation Metrics",
      "iframe_height": 900
    }
  ]
}