add vis to docs

Author: lukagerlach

README.md

@@ -72,6 +72,10 @@ python data/pybbbc_loader.py
 
 ## Usage
 
+### Notebooks
+
+For a very convenient way to check out our data, results and also training your own model, we created some Notebooks. you can find them in the notebooks folder. Else, you can just use the scripts that are described below:
+
 ### Training Models
 
 **Train vanilla SimCLR:**

docs/source/_static/plots/accuracy_comparison_cosine.png

@@ -28,8 +28,8 @@
 ## Visualization
 
 ```{eval-rst}
-.. automodule:: evaluation.visualize_embeddings
+.. automodule:: evaluation.visualization.visualize_embeddings
    :members:
    :undoc-members:
    :show-inheritance:
-```
+```

docs/source/_static/plots/confusion_matrices_cosine.png

@@ -30,7 +30,7 @@
 # -- Options for HTML output -------------------------------------------------
 
 html_theme = "sphinx_book_theme"
-# html_static_path = ["_static"]  # Commented out - no custom static files needed
+html_static_path = ["_static"]
 
 # Sphinx Book Theme options
 html_theme_options = {

docs/source/_static/plots/similarity_comparison.png

@@ -21,7 +21,7 @@ api/index
 :maxdepth: 2
 :caption: Research:
 
-experiments
+results
 ```
 
 ```{toctree}

docs/source/_static/plots/tsne_comparison_base_resnet_simclr_vanilla_ws_neg_simclr_vanilla_simclr_ws_collapsed.png

@@ -0,0 +1,27 @@
+# Results
+
+This section documents experimental results and analysis for compound profiling using different self-supervised learning approaches.
+
+## Model Performance Comparison
+
+The following plot shows the accuracy comparison between different models using cosine similarity as the distance metric:
+
+![Accuracy comparison using cosine similarity](_static/plots/accuracy_comparison_cosine.png)
+
+## Confusion Matrices
+
+Detailed confusion matrices for model evaluation. We show how treatments of each MOA are classified:
+
+![Confusion matrices with cosine similarity](_static/plots/confusion_matrices_cosine.png)
+
+## Similarity Analysis
+
+We compare similarity stats between same and different MOAs here:
+
+![Similarity comparison between models](_static/plots/similarity_comparison.png)
+
+## Dimensionality Reduction Visualization
+
+t-SNE visualization comparing treatment embeddings from different models (Base ResNet, SimCLR variants, and WS-DINO):
+
+![t-SNE comparison of model embeddings](_static/plots/tsne_comparison_base_resnet_simclr_vanilla_ws_neg_simclr_vanilla_simclr_ws_collapsed.png)

docs/source/api/evaluation.md

@@ -15,6 +15,8 @@
 from sklearn.metrics.pairwise import cosine_similarity
 import pandas as pd
 from collections import defaultdict
+from sklearn.neighbors import NearestNeighbors
+import math
 
 # Add parent directory to path to import evaluator
 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
@@ -125,13 +127,13 @@ def plot_confusion_matrices(model_names, data_root="/scratch/cv-course2025/group
         output_dir (str): Directory to save plots
         distance_measure (str): Distance measure for evaluation
     """
-    from sklearn.neighbors import NearestNeighbors
-    
+
     n_models = len(model_names)
-    fig, axes = plt.subplots(1, n_models, figsize=(6 * n_models, 5))
-    if n_models == 1:
-        axes = [axes]
-    
+    n_cols = 2
+    n_rows = math.ceil(n_models / n_cols)
+    fig, axes = plt.subplots(n_rows, n_cols, figsize=(6 * n_cols, 5 * n_rows))
+    axes = axes.flatten()  # Flatten in case of single row
+
     for i, model_name in enumerate(model_names):
         try:
             print(f"Creating confusion matrix for {model_name}...")
@@ -187,7 +189,11 @@ def plot_confusion_matrices(model_names, data_root="/scratch/cv-course2025/group
             ax.text(0.5, 0.5, f'Error: {str(e)}', transform=ax.transAxes, 
                    ha='center', va='center', fontsize=12)
             ax.set_title(f'{model_name} - Error')
-    
+
+    # Hide any unused subplots
+    for j in range(len(model_names), len(axes)):
+        fig.delaxes(axes[j])
+
     plt.suptitle(f'Confusion Matrices ({distance_measure} distance)', fontsize=16, fontweight='bold')
     plt.tight_layout()