diff --git a/IDEAS.md b/IDEAS.md
index c41d4dd..e9e0382 100644
--- a/IDEAS.md
+++ b/IDEAS.md
@@ -50,4 +50,14 @@
 📅 **Date:** 09/2026<br>
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
+---
+❓ **Question:** Is reidentification possible from simple embedding comparison ?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
+❓ **Question:** Can data registration and alignment be based on foundational features?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
 ---
\ No newline at end of file
diff --git a/batching/IDEAS.md b/batching/IDEAS.md
index 8a69128..bdaa742 100644
--- a/batching/IDEAS.md
+++ b/batching/IDEAS.md
@@ -5,4 +5,9 @@
 📅 **Date:** 08/2026<br>
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
+---
+❓ **Question:** Assess the difference between batches to analyse/justify gradient variations <br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
 ---
\ No newline at end of file
diff --git a/data_selection/IDEAS.md b/data_selection/IDEAS.md
index 3b2b9fa..c964f07 100644
--- a/data_selection/IDEAS.md
+++ b/data_selection/IDEAS.md
@@ -51,3 +51,13 @@
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
 ---
+❓ **Question:** Data required to optimize the performance depends on the training status, would it be possible to know this per task in advance ?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
+❓ **Question:** Use pretrained embeddings to select the best samples for test time in context injection<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
diff --git a/drift/IDEAS.md b/drift/IDEAS.md
index 0ecb6ab..75ac6a9 100644
--- a/drift/IDEAS.md
+++ b/drift/IDEAS.md
@@ -10,4 +10,9 @@
 📅 **Date:** 08/2026<br>
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
+---
+❓ **Question:** Does clustering help detect data drift by comparing embedding distributions?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
 ---
\ No newline at end of file
diff --git a/losses/IDEAS.md b/losses/IDEAS.md
index 69bf748..d40b5b8 100644
--- a/losses/IDEAS.md
+++ b/losses/IDEAS.md
@@ -20,4 +20,9 @@
 📅 **Date:** 08/2026<br>
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
+---
+❓ **Question:** Weight the loss contribution based on the embeddings of the full training set<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
 ---
\ No newline at end of file
diff --git a/model_design/IDEAS.md b/model_design/IDEAS.md
index ab20d84..578af38 100644
--- a/model_design/IDEAS.md
+++ b/model_design/IDEAS.md
@@ -66,3 +66,18 @@
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
 ---
+❓ **Question:** Explain model decisions by comparing embeddings with the training set<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
+❓ **Question:** Is it possible to detect failure mode by comparing model performance between clusters?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
+❓ **Question:** Find similar datasets from embedding similarity and reuse public studies to select best hyperparameters<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
+---
diff --git a/out_of_distribution/IDEAS.md b/out_of_distribution/IDEAS.md
index 3589152..939aec6 100644
--- a/out_of_distribution/IDEAS.md
+++ b/out_of_distribution/IDEAS.md
@@ -5,4 +5,9 @@
 📅 **Date:** 08/2026<br>
 👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
 🧭 **Context:** N/A<br>
+---
+❓ **Question:** Is it possible to detect out of distribution samples by comparing predictions in the same semantic cluster?<br>
+📅 **Date:** 10/2026<br>
+👤 **Author:** [Yann Chéné Leroy](https://github.com/Yann-CV)<br>
+🧭 **Context:** N/A<br>
 ---
\ No newline at end of file