From eeaa6776b1449b4226d1e04d1b9823316b08cc9d Mon Sep 17 00:00:00 2001
From: Alejandro Moreo <alejandro.moreo@isti.cnr.it>
Date: Thu, 20 Aug 2026 18:11:07 +0200
Subject: [PATCH] updated manuals

---
 CHANGE_LOG.txt                 |  7 +++++++
 TODO.txt                       | 14 --------------
 docs/source/manuals/methods.md | 31 ++++++++++++++++++++++++++++++-
 3 files changed, 37 insertions(+), 15 deletions(-)

diff --git a/CHANGE_LOG.txt b/CHANGE_LOG.txt
index a384d39..a364696 100644
--- a/CHANGE_LOG.txt
+++ b/CHANGE_LOG.txt
@@ -1,3 +1,10 @@
+Change Log 0.2.2
+-----------------
+
+- Added HistNetQ, based on the original implementation https://github.com/pglez84/histnetq
+
+- Minor fixes
+
 Change Log 0.2.1
 -----------------
 
diff --git a/TODO.txt b/TODO.txt
index f522a47..17604ab 100644
--- a/TODO.txt
+++ b/TODO.txt
@@ -1,14 +1,7 @@
-Adapt examples; remaining: example 4-onwards
-not working: 15 (qunfold)
-
-Solve the warnings issue; right now there is a warning ignore in method/__init__.py:
-
 Add 'platt' to calib options in EMQ?
 
 Allow n_prevpoints in APP to be specified by a user-defined grid?
 
-Update READMEs, wiki, & examples for new fit-predict interface
-
 Add the fix suggested by Alexander:
 
 For a more general application, I would maybe first establish a per-class threshold value of plausible prevalence
@@ -20,15 +13,8 @@ scale each value by per-class thresholds, i.e., [0.33*0.1, 0.33*1, 0.33*1]/sum.
 - This affects LabelledCollection
 - This functionality should be accessible via sampling protocols and evaluation functions
 
-Solve the pre-trained classifier issues. An example is the coptic-codes script I did, which needed a mock_lr to
-work for having access to classes_; think also the case in which the precomputed outputs are already generated
-as in the unifying problems code.
-
-
-- [TODO] document confidence in manuals
 - [TODO] Test the return_type="index" in protocols and finish the "distributing_samples.py" example
 - [TODO] add ensemble methods SC-MQ, MC-SQ, MC-MQ
-- [TODO] add HistNetQ
 - [TODO] add CDE-iteration and Bayes-CDE methods
 - [TODO] add Friedman's method and DeBias
 - [TODO] check ignore warning stuff
diff --git a/docs/source/manuals/methods.md b/docs/source/manuals/methods.md
index 4d7d75e..b721957 100644
--- a/docs/source/manuals/methods.md
+++ b/docs/source/manuals/methods.md
@@ -851,7 +851,36 @@ model.fit(*dataset.training.Xy)
 estim_prevalence = model.predict(dataset.test.X)
 ```
 
-(confidence-regions-for-class-prevalence-estimation)=
+### HistNetQ
+
+QuaPy offers an implementation of HistNetQ, a deep learning model based on a differentiable
+histogram representation, presented in:
+
+[_Pérez-Mon, O., Moreo, A., del Coz, J.J., & González, P. (2025).
+Quantification using permutation-invariant networks based on histograms.
+Neural Computing and Applications, 37(5), 3505-3520._](https://doi.org/10.1007/s00521-024-10721-1)
+
+This model requires `torch` to be installed. Like QuaNet, HistNetQ is trained end-to-end on
+samples ("bags") of known prevalence rather than on individually labeled instances; unlike QuaNet,
+it requires no classifier at all, only an optional feature extraction module (a plain identity
+module is used by default, for already-vectorized data).
+
+```python
+import quapy as qp
+from quapy.method.meta import HistNetQ
+
+dataset = qp.datasets.fetch_UCIBinaryDataset('haberman')
+
+model = HistNetQ(bag_size=100, device='cpu')
+model.fit(*dataset.training.Xy)
+estim_prevalence = model.predict(dataset.test.X)
+```
+
+HistNetQ can alternatively be trained directly from a protocol that already provides the training
+samples (e.g., when only bag-level prevalence values are available), via the `fit_from_samples`
+method; see the API documentation for further details.
+
+
 ## Quantifiers with Uncertainty Quantification
 
 _(New in v0.2.0!)_ Some quantification methods go beyond providing a single point estimate of class prevalence values and also produce confidence regions, which characterize the uncertainty around the point estimate. In QuaPy, two such families are currently implemented: bootstrap methods and Bayesian methods.