python-ldl 0.0.3__tar.gz → 0.1.0__tar.gz

{python_ldl-0.0.3 → python_ldl-0.1.0}/LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2024 SpriteMisaka
+Copyright (c) 2024-2026 SpriteMisaka
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

{python_ldl-0.0.3/python_ldl.egg-info → python_ldl-0.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: python-ldl
-Version: 0.0.3
+Version: 0.1.0
 Summary: Label distribution learning (LDL) and label enhancement (LE) toolkit implemented in python.
 Home-page: https://github.com/SpriteMisaka/PyLDL
 Author: SpriteMisaka
@@ -12,55 +12,101 @@ Requires-Python: >=3
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: matplotlib
+Requires-Dist: keras>=3
+Requires-Dist: numba
 Requires-Dist: numpy
 Requires-Dist: qpsolvers
 Requires-Dist: quadprog
 Requires-Dist: scikit-fuzzy
 Requires-Dist: scikit-learn
 Requires-Dist: scipy
-Requires-Dist: tensorflow
-Requires-Dist: tensorflow-probability
+Provides-Extra: tensorflow
+Requires-Dist: tensorflow; extra == "tensorflow"
+Requires-Dist: tensorflow-probability[tf]; extra == "tensorflow"
+Provides-Extra: torch
+Requires-Dist: torch; extra == "torch"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # PyLDL
 
 Label distribution learning (LDL) and label enhancement (LE) toolkit implemented in python, including:
 
 + LDL algorithms:
-  + ([Geng, Yin, and Zhou 2013](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2013.pdf))[*TPAMI*]: `CPNN`$^1$.
-  + ([Geng and Hou 2015](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2015.pdf))[*IJCAI*]: `LDSVR`.
-  + ⭐([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf))[*TKDE*]: `SA_BFGS`, `SA_IIS`, `AA_KNN`, `AA_BP`, `PT_Bayes`, and `PT_SVM`.
-  + ([Yang, Sun, and Sun 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/yang2017.pdf))[*AAAI*]: `BCPNN` and `ACPNN`.
-  + ([Xu and Zhou 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2017.pdf))[*IJCAI*]: `IncomLDL`$^2$.
-  + ([Shen et al. 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shen2017.pdf))[*NeurIPS*]: `LDLF`.
-  + ([Wang and Geng 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2019.pdf))[*IJCAI*]: `LDL4C`$^3$.
-  + ([Shen et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shen2020.pdf))[*南京理工大学学报* (Chinese)]: `AdaBoostLDL`.
-  + ([González et al. 2021a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/gonz%C3%A1lez2021a.pdf))[*Inf. Sci.*]: `SSG_LDL`$^4$.
-  + ([González et al. 2021b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/gonz%C3%A1lez2021b.pdf))[*Inf. Fusion*]: `DF_LDL`.
-  + ([Wang and Geng 2021a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2021a.pdf))[*IJCAI*]: `LDL_HR`$^3$.
-  + ([Wang and Geng 2021b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2021b.pdf))[*ICML*]: `LDLM`$^3$.
-  + ([Jia et al. 2021](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2021.pdf))[*TKDE*]: `LDL_SCL`.
-  + ([Jia et al. 2023a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2023a.pdf))[*TKDE*]: `LDL_LRR`.
-  + ([Jia et al. 2023b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2023b.pdf))[*TNNLS*]: `LDL_DPA`.
-  + ([Wen et al. 2023](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wen2023.pdf))[*ICCV*]: `CAD`$^1$, `QFD2`$^1$, and `CJS`$^1$.
+  + ([Geng, Yin, and Zhou 2013](https://doi.org/10.1109/tpami.2013.51)) [*TPAMI*]: `CPNN`$^1$.
+  + ([Geng and Hou 2015](https://www.ijcai.org/Abstract/15/494)) [*IJCAI*]: `LDSVR`.
+  + ⭐([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) [*TKDE*]: `SA_BFGS`, `SA_IIS`, `AA_KNN`, `AA_BP`, `PT_Bayes`, and `PT_SVM`.
+  + ([Yang, Sun, and Sun 2017](https://doi.org/10.1609/aaai.v31i1.10485)) [*AAAI*]: `BCPNN` and `ACPNN`.
+  + ([Xu and Zhou 2017](https://doi.org/10.24963/ijcai.2017/443)) [*IJCAI*]: `IncomLDL`$^2$.
+  + ([Shen et al. 2017](https://papers.nips.cc/paper_files/paper/2017/hash/6e2713a6efee97bacb63e52c54f0ada0-Abstract.html)) [*NeurIPS*]: `LDLF`.
+  + ([Zhao and Zhou 2018](https://doi.org/10.1609/aaai.v32i1.11609)) [*AAAI*]: `LALOT`.
+  + ([Jia et al. 2018](https://doi.org/10.1609/aaai.v32i1.11664)) [*AAAI*]: `LDLLC`$^\dagger$.
+  + ([Chen et al. 2018](https://doi.org/10.1016/j.neucom.2018.09.002)) [*Neurocomputing*]: `StructTree` and `StructRF`.
+  + ([Ren et al. 2019a](https://doi.org/10.24963/ijcai.2019/460)) [*IJCAI*]: `LDLSF`.
+  + ([Ren et al. 2019b](https://doi.org/10.24963/ijcai.2019/461)) [*IJCAI*]: `LDL_LCLR`$^\dagger$.
+  + ([Wang and Geng 2019](https://doi.org/10.24963/ijcai.2019/515)) [*IJCAI*]: `LDL4C`$^{3\dagger}$.
+  + ([González et al. 2021a](https://doi.org/10.1016/j.ins.2020.07.071)) [*Inf. Sci.*]: `SSG_LDL`$^{4}$.
+  + ([González et al. 2021b](https://doi.org/10.1016/j.inffus.2020.08.024)) [*Inf. Fusion*]: `DF_LDL`.
+  + ([Wang and Geng 2021a](https://doi.org/10.24963/ijcai.2021/426)) [*IJCAI*]: `LDL_HR`$^{3^\dagger}$.
+  + ([Wang and Geng 2021b](https://proceedings.mlr.press/v139/wang21h.html)) [*ICML*]: `LDLM`$^{3^\dagger}$.
+  + ([Jia et al. 2021](https://doi.org/10.1109/TKDE.2019.2943337)) [*TKDE*]: `LDL_SCL`.
+  + ([Liu  et al. 2021](https://doi.org/10.1016/j.knosys.2020.106690)) [*KBS*]: `BD_LDL`.
+  + ([Żychowski and Mańdziuk 2021](https://doi.org/10.1016/j.asoc.2021.107585)) [*Appl. Soft Comput.*]: `Duo_LDL`.
+  + ([Li et al. 2022](https://doi.org/10.1109/CVPR52688.2022.01986)) [*CVPR*]: `unimodal_loss`$^1$ and `concentrated_loss`$^1$.
+  + ([Jia et al. 2023a](https://doi.org/10.1109/TKDE.2021.3099294)) [*TKDE*]: `LDL_LRR`$^\dagger$.
+  + ([Jia et al. 2023b](https://doi.org/10.1109/TNNLS.2023.3258976)) [*TNNLS*]: `LDL_DPA`$^\dagger$.
+  + ([Wen et al. 2023](https://doi.org/10.1109/ICCV51070.2023.02146)) [*ICCV*]: `cad`$^1$, `qfd2`$^1$, and `cjs`$^1$.
+  + ([Li and Chen 2024](https://doi.org/10.24963/ijcai.2024/494)) [*IJCAI*]: `WInLDL`$^2$.
+  + ([Kou et al. 2024](https://doi.org/10.24963/ijcai.2024/478)) [*IJCAI*]: `TLRLDL`$^\dagger$ and `TKLRLDL`$^\dagger$.
+  + ([Lin et al. 2024](https://doi.org/10.1109/TBDATA.2023.3338023)) [*TBD*]: `LDL_HVLC`$^\dagger$.
+  + ([Wu, Li, and Jia 2025](https://doi.org/10.1109/TBDATA.2024.3442562)) [*TBD*]: `LDL_DA`$^5$.
+  + ([Wang et al. 2025](https://doi.org/10.1016/j.patcog.2024.111006)) [*Pattern Recognit.*]: `RKNN_LDL`$^\dagger$.
+  + ([Tan et al. 2025](https://doi.org/10.1016/j.knosys.2025.113666)) [*KBS*]: `RG4LDL`.
+  + ([Wu, Li, and Jia 2025](https://icml.cc/virtual/2025/poster/44379)) [*ICML*]: `S_LRR`, `S_SCL`, `S_KLD`, `S_CJS` and `S_QFD2`.
+  + ([Li et al. 2025](https://icml.cc/virtual/2025/poster/46395)) [*ICML*]: `Delta_LDL`.
+  + ... and many more LDL algorithms are included! Discover them in the `pyldl.algorithms` module! 🚀
 + LE algorithms:
-  + ([Xu, Liu, and Geng 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2019.pdf))[*TKDE*]: `FCM`, `KM`, `LP`, `ML`, and `GLLE`.
-  + ([Xu et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2020.pdf))[*ICML*]: `LEVI`.
-  + ([Zheng, Zhu, and Tang 2023](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/zheng2023.pdf))[*CVPR*]: `LIBLE`.
+  + ([Xu, Liu, and Geng 2019](https://doi.org/10.1109/TKDE.2019.2947040)) [*TKDE*]: `FCM`, `KM`, `LP`, `ML`, and `GLLE`.
+  + ([Xu et al. 2020](https://proceedings.mlr.press/v119/xu20g.html)) [*ICML*]: `LEVI`.
+  + ([Zheng, Zhu, and Tang 2023](https://doi.org/10.1109/CVPR52729.2023.00724)) [*CVPR*]: `LIBLE`.
+  + ([Wang et al. 2023](https://doi.org/10.24963/ijcai.2023/484)) [*IJCAI*]: `ConLE`.
 + LDL metrics: `chebyshev`, `clark`, `canberra`, `kl_divergence`, `cosine`, `intersection`, etc.
 + Structured LDL datasets: *Human_Gene*, *Movie*, *Natural_Scene*, *s-BU_3DFE*, *s-JAFFE*, *Yeast*, etc.
 + LDL applications:
-  + Facial emotion recognition (supported datasets: [*JAFFE*](https://zenodo.org/records/3451524))
-  + ([Shirani et al. 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shirani2019.pdf))[*ACL*]: Emphasis selection (supported datasets: [*SemEval2020*](https://github.com/RiTUAL-UH/SemEval2020_Task10_Emphasis_Selection); pre-trained GloVe embeddings can be downloaded [here](https://nlp.stanford.edu/projects/glove/)).
-  + ([Wu et al. 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wu2019.pdf))[*ICCV*]: Lesion counting (supported datasets: [*ACNE04*](https://drive.google.com/drive/folders/18yJcHXhzOv7H89t-Lda6phheAicLqMuZ)).
-  + ([Chen et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/chen2020.pdf))[*CVPR*]: Facial emotion recognition with auxiliary label space graphs (supported datasets: [*CK+*](https://www.jeffcohn.net/Resources/); OpenFace can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/releases), and the required models can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/wiki/Model-download)).
+  + Facial emotion recognition (supported datasets: [*JAFFE*](https://zenodo.org/records/3451524) and [*BU-3DFE*](https://www.cs.binghamton.edu/~lijun/Research/3DFE/3DFE_Analysis.html)).
+  + ([Shirani et al. 2019](https://doi.org/10.18653/v1/P19-1112)) [*ACL*]: Emphasis selection (supported datasets: [*SemEval2020*](https://github.com/RiTUAL-UH/SemEval2020_Task10_Emphasis_Selection); pre-trained GloVe embeddings can be downloaded [here](https://nlp.stanford.edu/projects/glove/)).
+  + ([Wu et al. 2019](https://doi.org/10.1109/ICCV.2019.01074)) [*ICCV*]: Lesion counting (supported datasets: [*ACNE04*](https://drive.google.com/drive/folders/18yJcHXhzOv7H89t-Lda6phheAicLqMuZ)).
+  + ([Chen et al. 2020](https://doi.org/10.1109/CVPR42600.2020.01400)) [*CVPR*]: Facial emotion recognition with auxiliary label space graphs (supported datasets: [*CK+*](https://www.jeffcohn.net/Resources/); OpenFace can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/releases), and the required models can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/wiki/Model-download)).
 
 > $^1$ Technically, these methods are only suitable for totally ordered labels.
 >
-> $^2$ These are algorithms for incomplete LDL, so you should use `pyldl.utils.random_missing` to generate the missing label distribution matrix and the corresponding mask matrix in the experiments.
+> $^2$ These are algorithms for incomplete LDL, so you should use `pyldl.utils.random_missing` to generate the missing label distribution matrix and the corresponding mask matrix in the experiments. [Here](https://github.com/SpriteMisaka/PyLDL/blob/main/demo/incomplete_settings.ipynb) is a demo on using the incomplete LDL algorithms.
 >
 > $^3$ These are LDL classifiers, so you should use `predict_proba` to get label distributions and `predict` to get predicted labels.
 >
 > $^4$ These are oversampling algorithms for LDL, therefore you should use `fit_transform` to generate synthetic samples.
+>
+> $^5$ To use domain adaptation methods for LDL, you need to provide the source domain data via parameters `sX` and `sy` of the `fit` method. [Here](https://github.com/SpriteMisaka/PyLDL/blob/main/demo/domain_adaptation.ipynb) is a demo on domain adaptation for LDL.
+
+> $^\dagger$ These methods involve imposing constraints on model parameters, like regularization. Therefore, it is recommended to carefully tune the hyperparameters and apply feature preprocessing techniques like `StandardScaler` or `MinMaxScaler` before conducting experiments to achieve the expected performance.
+
+## ❗ News
+
+- **0.1.0**: PyLDL now supports both TensorFlow and PyTorch backends through Keras 3. You can switch between the two backends by setting the `KERAS_BACKEND` environment variable:
+
+```python
+import os
+os.environ["KERAS_BACKEND"] = "torch"
+```
 
 ## Installation
 
@@ -88,14 +134,14 @@ from pyldl.metrics import score
 from sklearn.model_selection import train_test_split
 
 dataset_name = 'SJAFFE'
-X, y = load_dataset(dataset_name)
-X_train, X_test, y_train, y_test = train_test_split(X, y)
+X, D = load_dataset(dataset_name)
+X_train, X_test, D_train, D_test = train_test_split(X, D)
 
 model = SA_BFGS()
-model.fit(X_train, y_train)
+model.fit(X_train, D_train)
 
-y_pred = model.predict(X_test)
-print(score(y_test, y_pred))
+D_pred = model.predict(X_test)
+print(score(D_test, D_pred))
 ```
 
 For those who would like to use the original implementation:
@@ -112,7 +158,7 @@ Now, you can load the original implementation of the method, e.g.:
 from pyldl.matlab_algorithms import SA_IIS
 ```
 
-You can visualize the performance of any model on the artificial dataset ([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf)) with the `pyldl.utils.plot_artificial` function, e.g.:
+You can visualize the performance of any model on the artificial dataset ([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) with the `pyldl.utils.plot_artificial` function, e.g.:
 
 ```python
 from pyldl.algorithms import LDSVR, SA_BFGS, SA_IIS, AA_KNN, PT_Bayes, GLLE, LIBLE
@@ -143,7 +189,7 @@ The output images are as follows.
 | :----------------------------------------------------------: | :----------------------------------------------------------: |
 |                            `GLLE`                            |                           `LIBLE`                            |
 
-Enjoy! :)
+Refer to the [docs](https://spritemisaka.github.io/PyLDL/) for more detailed information. Enjoy! :)
 
 ## Experiments
 
@@ -160,7 +206,7 @@ Results of ours are as follows.
 | PT-Bayes  |   .116 ± .011   |   .425 ± .031   |   .874 ± .064   |   .073 ± .012   |   .932 ± .011   |   .850 ± .012   |
 |  PT-SVM   |   .117 ± .012   |   .422 ± .027   |   .875 ± .057   |   .072 ± .011   |   .932 ± .011   |   .850 ± .011   |
 
-Results of the original MATLAB implementation ([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf)) are as follows.
+Results of the original MATLAB implementation ([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) are as follows.
 
 | Algorithm |    Cheby.(↓)    |    Clark(↓)     |     Can.(↓)     |     K-L(↓)      |     Cos.(↑)     |     Int.(↑)     |
 | :-------: | :-------------: | :-------------: | :-------------: | :-------------: | :-------------: | :-------------: |
@@ -174,14 +220,26 @@ Results of the original MATLAB implementation ([Geng 2016](https://github.com/Sp
 ## Requirements
 
 ```
-matplotlib>=3.6.1
-numpy>=1.22.3
-qpsolvers>=4.0.0
-quadprog>=0.1.11
-scikit-fuzzy>=0.4.2
-scikit-learn>=1.0.2
-scipy>=1.8.0
-tensorflow>=2.8.0
-tensorflow-probability>=0.16.0
+Cython
+matplotlib
+numba
+numpy
+qpsolvers
+quadprog
+scikit-fuzzy
+scikit-learn
+scipy
 ```
 
+If you use TensorFlow backend, you also need to install:
+
+```
+tensorflow
+tensorflow-probability
+```
+
+If you use PyTorch backend, you also need to install:
+
+```
+torch
+```

{python_ldl-0.0.3 → python_ldl-0.1.0}/README.md

@@ -3,41 +3,71 @@
 Label distribution learning (LDL) and label enhancement (LE) toolkit implemented in python, including:
 
 + LDL algorithms:
-  + ([Geng, Yin, and Zhou 2013](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2013.pdf))[*TPAMI*]: `CPNN`$^1$.
-  + ([Geng and Hou 2015](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2015.pdf))[*IJCAI*]: `LDSVR`.
-  + ⭐([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf))[*TKDE*]: `SA_BFGS`, `SA_IIS`, `AA_KNN`, `AA_BP`, `PT_Bayes`, and `PT_SVM`.
-  + ([Yang, Sun, and Sun 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/yang2017.pdf))[*AAAI*]: `BCPNN` and `ACPNN`.
-  + ([Xu and Zhou 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2017.pdf))[*IJCAI*]: `IncomLDL`$^2$.
-  + ([Shen et al. 2017](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shen2017.pdf))[*NeurIPS*]: `LDLF`.
-  + ([Wang and Geng 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2019.pdf))[*IJCAI*]: `LDL4C`$^3$.
-  + ([Shen et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shen2020.pdf))[*南京理工大学学报* (Chinese)]: `AdaBoostLDL`.
-  + ([González et al. 2021a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/gonz%C3%A1lez2021a.pdf))[*Inf. Sci.*]: `SSG_LDL`$^4$.
-  + ([González et al. 2021b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/gonz%C3%A1lez2021b.pdf))[*Inf. Fusion*]: `DF_LDL`.
-  + ([Wang and Geng 2021a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2021a.pdf))[*IJCAI*]: `LDL_HR`$^3$.
-  + ([Wang and Geng 2021b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wang2021b.pdf))[*ICML*]: `LDLM`$^3$.
-  + ([Jia et al. 2021](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2021.pdf))[*TKDE*]: `LDL_SCL`.
-  + ([Jia et al. 2023a](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2023a.pdf))[*TKDE*]: `LDL_LRR`.
-  + ([Jia et al. 2023b](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/jia2023b.pdf))[*TNNLS*]: `LDL_DPA`.
-  + ([Wen et al. 2023](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wen2023.pdf))[*ICCV*]: `CAD`$^1$, `QFD2`$^1$, and `CJS`$^1$.
+  + ([Geng, Yin, and Zhou 2013](https://doi.org/10.1109/tpami.2013.51)) [*TPAMI*]: `CPNN`$^1$.
+  + ([Geng and Hou 2015](https://www.ijcai.org/Abstract/15/494)) [*IJCAI*]: `LDSVR`.
+  + ⭐([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) [*TKDE*]: `SA_BFGS`, `SA_IIS`, `AA_KNN`, `AA_BP`, `PT_Bayes`, and `PT_SVM`.
+  + ([Yang, Sun, and Sun 2017](https://doi.org/10.1609/aaai.v31i1.10485)) [*AAAI*]: `BCPNN` and `ACPNN`.
+  + ([Xu and Zhou 2017](https://doi.org/10.24963/ijcai.2017/443)) [*IJCAI*]: `IncomLDL`$^2$.
+  + ([Shen et al. 2017](https://papers.nips.cc/paper_files/paper/2017/hash/6e2713a6efee97bacb63e52c54f0ada0-Abstract.html)) [*NeurIPS*]: `LDLF`.
+  + ([Zhao and Zhou 2018](https://doi.org/10.1609/aaai.v32i1.11609)) [*AAAI*]: `LALOT`.
+  + ([Jia et al. 2018](https://doi.org/10.1609/aaai.v32i1.11664)) [*AAAI*]: `LDLLC`$^\dagger$.
+  + ([Chen et al. 2018](https://doi.org/10.1016/j.neucom.2018.09.002)) [*Neurocomputing*]: `StructTree` and `StructRF`.
+  + ([Ren et al. 2019a](https://doi.org/10.24963/ijcai.2019/460)) [*IJCAI*]: `LDLSF`.
+  + ([Ren et al. 2019b](https://doi.org/10.24963/ijcai.2019/461)) [*IJCAI*]: `LDL_LCLR`$^\dagger$.
+  + ([Wang and Geng 2019](https://doi.org/10.24963/ijcai.2019/515)) [*IJCAI*]: `LDL4C`$^{3\dagger}$.
+  + ([González et al. 2021a](https://doi.org/10.1016/j.ins.2020.07.071)) [*Inf. Sci.*]: `SSG_LDL`$^{4}$.
+  + ([González et al. 2021b](https://doi.org/10.1016/j.inffus.2020.08.024)) [*Inf. Fusion*]: `DF_LDL`.
+  + ([Wang and Geng 2021a](https://doi.org/10.24963/ijcai.2021/426)) [*IJCAI*]: `LDL_HR`$^{3^\dagger}$.
+  + ([Wang and Geng 2021b](https://proceedings.mlr.press/v139/wang21h.html)) [*ICML*]: `LDLM`$^{3^\dagger}$.
+  + ([Jia et al. 2021](https://doi.org/10.1109/TKDE.2019.2943337)) [*TKDE*]: `LDL_SCL`.
+  + ([Liu  et al. 2021](https://doi.org/10.1016/j.knosys.2020.106690)) [*KBS*]: `BD_LDL`.
+  + ([Żychowski and Mańdziuk 2021](https://doi.org/10.1016/j.asoc.2021.107585)) [*Appl. Soft Comput.*]: `Duo_LDL`.
+  + ([Li et al. 2022](https://doi.org/10.1109/CVPR52688.2022.01986)) [*CVPR*]: `unimodal_loss`$^1$ and `concentrated_loss`$^1$.
+  + ([Jia et al. 2023a](https://doi.org/10.1109/TKDE.2021.3099294)) [*TKDE*]: `LDL_LRR`$^\dagger$.
+  + ([Jia et al. 2023b](https://doi.org/10.1109/TNNLS.2023.3258976)) [*TNNLS*]: `LDL_DPA`$^\dagger$.
+  + ([Wen et al. 2023](https://doi.org/10.1109/ICCV51070.2023.02146)) [*ICCV*]: `cad`$^1$, `qfd2`$^1$, and `cjs`$^1$.
+  + ([Li and Chen 2024](https://doi.org/10.24963/ijcai.2024/494)) [*IJCAI*]: `WInLDL`$^2$.
+  + ([Kou et al. 2024](https://doi.org/10.24963/ijcai.2024/478)) [*IJCAI*]: `TLRLDL`$^\dagger$ and `TKLRLDL`$^\dagger$.
+  + ([Lin et al. 2024](https://doi.org/10.1109/TBDATA.2023.3338023)) [*TBD*]: `LDL_HVLC`$^\dagger$.
+  + ([Wu, Li, and Jia 2025](https://doi.org/10.1109/TBDATA.2024.3442562)) [*TBD*]: `LDL_DA`$^5$.
+  + ([Wang et al. 2025](https://doi.org/10.1016/j.patcog.2024.111006)) [*Pattern Recognit.*]: `RKNN_LDL`$^\dagger$.
+  + ([Tan et al. 2025](https://doi.org/10.1016/j.knosys.2025.113666)) [*KBS*]: `RG4LDL`.
+  + ([Wu, Li, and Jia 2025](https://icml.cc/virtual/2025/poster/44379)) [*ICML*]: `S_LRR`, `S_SCL`, `S_KLD`, `S_CJS` and `S_QFD2`.
+  + ([Li et al. 2025](https://icml.cc/virtual/2025/poster/46395)) [*ICML*]: `Delta_LDL`.
+  + ... and many more LDL algorithms are included! Discover them in the `pyldl.algorithms` module! 🚀
 + LE algorithms:
-  + ([Xu, Liu, and Geng 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2019.pdf))[*TKDE*]: `FCM`, `KM`, `LP`, `ML`, and `GLLE`.
-  + ([Xu et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/xu2020.pdf))[*ICML*]: `LEVI`.
-  + ([Zheng, Zhu, and Tang 2023](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/zheng2023.pdf))[*CVPR*]: `LIBLE`.
+  + ([Xu, Liu, and Geng 2019](https://doi.org/10.1109/TKDE.2019.2947040)) [*TKDE*]: `FCM`, `KM`, `LP`, `ML`, and `GLLE`.
+  + ([Xu et al. 2020](https://proceedings.mlr.press/v119/xu20g.html)) [*ICML*]: `LEVI`.
+  + ([Zheng, Zhu, and Tang 2023](https://doi.org/10.1109/CVPR52729.2023.00724)) [*CVPR*]: `LIBLE`.
+  + ([Wang et al. 2023](https://doi.org/10.24963/ijcai.2023/484)) [*IJCAI*]: `ConLE`.
 + LDL metrics: `chebyshev`, `clark`, `canberra`, `kl_divergence`, `cosine`, `intersection`, etc.
 + Structured LDL datasets: *Human_Gene*, *Movie*, *Natural_Scene*, *s-BU_3DFE*, *s-JAFFE*, *Yeast*, etc.
 + LDL applications:
-  + Facial emotion recognition (supported datasets: [*JAFFE*](https://zenodo.org/records/3451524))
-  + ([Shirani et al. 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/shirani2019.pdf))[*ACL*]: Emphasis selection (supported datasets: [*SemEval2020*](https://github.com/RiTUAL-UH/SemEval2020_Task10_Emphasis_Selection); pre-trained GloVe embeddings can be downloaded [here](https://nlp.stanford.edu/projects/glove/)).
-  + ([Wu et al. 2019](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/wu2019.pdf))[*ICCV*]: Lesion counting (supported datasets: [*ACNE04*](https://drive.google.com/drive/folders/18yJcHXhzOv7H89t-Lda6phheAicLqMuZ)).
-  + ([Chen et al. 2020](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/chen2020.pdf))[*CVPR*]: Facial emotion recognition with auxiliary label space graphs (supported datasets: [*CK+*](https://www.jeffcohn.net/Resources/); OpenFace can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/releases), and the required models can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/wiki/Model-download)).
+  + Facial emotion recognition (supported datasets: [*JAFFE*](https://zenodo.org/records/3451524) and [*BU-3DFE*](https://www.cs.binghamton.edu/~lijun/Research/3DFE/3DFE_Analysis.html)).
+  + ([Shirani et al. 2019](https://doi.org/10.18653/v1/P19-1112)) [*ACL*]: Emphasis selection (supported datasets: [*SemEval2020*](https://github.com/RiTUAL-UH/SemEval2020_Task10_Emphasis_Selection); pre-trained GloVe embeddings can be downloaded [here](https://nlp.stanford.edu/projects/glove/)).
+  + ([Wu et al. 2019](https://doi.org/10.1109/ICCV.2019.01074)) [*ICCV*]: Lesion counting (supported datasets: [*ACNE04*](https://drive.google.com/drive/folders/18yJcHXhzOv7H89t-Lda6phheAicLqMuZ)).
+  + ([Chen et al. 2020](https://doi.org/10.1109/CVPR42600.2020.01400)) [*CVPR*]: Facial emotion recognition with auxiliary label space graphs (supported datasets: [*CK+*](https://www.jeffcohn.net/Resources/); OpenFace can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/releases), and the required models can be downloaded [here](https://github.com/TadasBaltrusaitis/OpenFace/wiki/Model-download)).
 
 > $^1$ Technically, these methods are only suitable for totally ordered labels.
 >
-> $^2$ These are algorithms for incomplete LDL, so you should use `pyldl.utils.random_missing` to generate the missing label distribution matrix and the corresponding mask matrix in the experiments.
+> $^2$ These are algorithms for incomplete LDL, so you should use `pyldl.utils.random_missing` to generate the missing label distribution matrix and the corresponding mask matrix in the experiments. [Here](https://github.com/SpriteMisaka/PyLDL/blob/main/demo/incomplete_settings.ipynb) is a demo on using the incomplete LDL algorithms.
 >
 > $^3$ These are LDL classifiers, so you should use `predict_proba` to get label distributions and `predict` to get predicted labels.
 >
 > $^4$ These are oversampling algorithms for LDL, therefore you should use `fit_transform` to generate synthetic samples.
+>
+> $^5$ To use domain adaptation methods for LDL, you need to provide the source domain data via parameters `sX` and `sy` of the `fit` method. [Here](https://github.com/SpriteMisaka/PyLDL/blob/main/demo/domain_adaptation.ipynb) is a demo on domain adaptation for LDL.
+
+> $^\dagger$ These methods involve imposing constraints on model parameters, like regularization. Therefore, it is recommended to carefully tune the hyperparameters and apply feature preprocessing techniques like `StandardScaler` or `MinMaxScaler` before conducting experiments to achieve the expected performance.
+
+## ❗ News
+
+- **0.1.0**: PyLDL now supports both TensorFlow and PyTorch backends through Keras 3. You can switch between the two backends by setting the `KERAS_BACKEND` environment variable:
+
+```python
+import os
+os.environ["KERAS_BACKEND"] = "torch"
+```
 
 ## Installation
 
@@ -65,14 +95,14 @@ from pyldl.metrics import score
 from sklearn.model_selection import train_test_split
 
 dataset_name = 'SJAFFE'
-X, y = load_dataset(dataset_name)
-X_train, X_test, y_train, y_test = train_test_split(X, y)
+X, D = load_dataset(dataset_name)
+X_train, X_test, D_train, D_test = train_test_split(X, D)
 
 model = SA_BFGS()
-model.fit(X_train, y_train)
+model.fit(X_train, D_train)
 
-y_pred = model.predict(X_test)
-print(score(y_test, y_pred))
+D_pred = model.predict(X_test)
+print(score(D_test, D_pred))
 ```
 
 For those who would like to use the original implementation:
@@ -89,7 +119,7 @@ Now, you can load the original implementation of the method, e.g.:
 from pyldl.matlab_algorithms import SA_IIS
 ```
 
-You can visualize the performance of any model on the artificial dataset ([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf)) with the `pyldl.utils.plot_artificial` function, e.g.:
+You can visualize the performance of any model on the artificial dataset ([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) with the `pyldl.utils.plot_artificial` function, e.g.:
 
 ```python
 from pyldl.algorithms import LDSVR, SA_BFGS, SA_IIS, AA_KNN, PT_Bayes, GLLE, LIBLE
@@ -120,7 +150,7 @@ The output images are as follows.
 | :----------------------------------------------------------: | :----------------------------------------------------------: |
 |                            `GLLE`                            |                           `LIBLE`                            |
 
-Enjoy! :)
+Refer to the [docs](https://spritemisaka.github.io/PyLDL/) for more detailed information. Enjoy! :)
 
 ## Experiments
 
@@ -137,7 +167,7 @@ Results of ours are as follows.
 | PT-Bayes  |   .116 ± .011   |   .425 ± .031   |   .874 ± .064   |   .073 ± .012   |   .932 ± .011   |   .850 ± .012   |
 |  PT-SVM   |   .117 ± .012   |   .422 ± .027   |   .875 ± .057   |   .072 ± .011   |   .932 ± .011   |   .850 ± .011   |
 
-Results of the original MATLAB implementation ([Geng 2016](https://github.com/SpriteMisaka/PyLDL/blob/main/bibliography/geng2016.pdf)) are as follows.
+Results of the original MATLAB implementation ([Geng 2016](https://doi.org/10.1109/TKDE.2016.2545658)) are as follows.
 
 | Algorithm |    Cheby.(↓)    |    Clark(↓)     |     Can.(↓)     |     K-L(↓)      |     Cos.(↑)     |     Int.(↑)     |
 | :-------: | :-------------: | :-------------: | :-------------: | :-------------: | :-------------: | :-------------: |
@@ -151,14 +181,26 @@ Results of the original MATLAB implementation ([Geng 2016](https://github.com/Sp
 ## Requirements
 
 ```
-matplotlib>=3.6.1
-numpy>=1.22.3
-qpsolvers>=4.0.0
-quadprog>=0.1.11
-scikit-fuzzy>=0.4.2
-scikit-learn>=1.0.2
-scipy>=1.8.0
-tensorflow>=2.8.0
-tensorflow-probability>=0.16.0
+Cython
+matplotlib
+numba
+numpy
+qpsolvers
+quadprog
+scikit-fuzzy
+scikit-learn
+scipy
 ```
 
+If you use TensorFlow backend, you also need to install:
+
+```
+tensorflow
+tensorflow-probability
+```
+
+If you use PyTorch backend, you also need to install:
+
+```
+torch
+```

python_ldl-0.1.0/pyldl/algorithms/__init__.py

@@ -0,0 +1,198 @@
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from ._problem_transformation import *
+    from ._algorithm_adaptation import *
+    from ._specialized_algorithms import *
+
+    from ._incomplete import *
+    from ._classifier import *
+    from ._ensemble import *
+
+    from ._bp import *
+    from ._cpnn import *
+    from ._duo_ldl import *
+    from ._ldlf import *
+    from ._ldllc import *
+    from ._ldlsf import *
+    from ._ldl_lclr import *
+    from ._ldl_scl import *
+    from ._ldl_lrr import *
+    from ._ldl_dpa import *
+    from ._lrldl import *
+    from ._ldl_hvlc import *
+    from ._rknn_ldl import *
+    from ._s_ldl import *
+    from ._delta_ldl import *
+    from ._snefy_ldl import *
+    from ._ldl_dvs import *
+    from ._ldl_dpm import *
+
+    from ._ssg_ldl import *
+
+    from ._label_enhancement import *
+
+    from ._ldl_da import *
+
+
+_LDL_MODULE_MAP = {
+    # -------------------- 2026 --------------------
+    "LDL_DVS": "._ldl_dvs",
+    "LDL_DPM": "._ldl_dpm",
+    # -------------------- 2025 --------------------
+    "RG4LDL": "._ensemble",
+    "RKNN_LDL": "._rknn_ldl",
+    "SNEFY_LDL": "._snefy_ldl",
+    **dict.fromkeys(
+        [
+            "_S_LDL",
+            "S_LRR",
+            "S_SCL",
+            "S_KLD",
+            "S_CJS",
+            "S_QFD2",
+            "Shallow_S_LDL",
+        ],
+        "._s_ldl"
+    ),
+    "Delta_LDL": "._delta_ldl",
+# -------------------- 2024 --------------------
+    "LDL_HVLC": "._ldl_hvlc",
+    **dict.fromkeys(
+        [
+            "_LRLDL",
+            "TKLRLDL",
+            "TLRLDL",
+        ],
+        "._lrldl"
+    ),
+    "LDL_DPA": "._ldl_dpa",
+    # -------------------- 2023 --------------------
+    "LDL_LRR": "._ldl_lrr",
+    # -------------------- 2021 --------------------
+    "DF_LDL": "._ensemble",
+    "LDL_SCL": "._ldl_scl",
+    "Duo_LDL": "._duo_ldl",
+    "BD_LDL": "._algorithm_adaptation",
+    # -------------------- 2019 --------------------
+    "LDL_LCLR": "._ldl_lclr",
+    "LDLSF": "._ldlsf",
+    # -------------------- 2018 --------------------
+    "LDLLC": "._ldllc",
+    "LALOT": "._specialized_algorithms",
+    "StructRF": "._ensemble",
+    # -------------------- 2017 --------------------
+    "BCPNN": "._cpnn",
+    "ACPNN": "._cpnn",
+    "LDLF": "._ldlf",
+    # -------------------- 2016 --------------------
+    "LDLogitBoost": "._ensemble",
+    **dict.fromkeys(
+        [
+            "_SA",
+            "SA_BFGS",
+            "SA_IIS",
+        ],
+        "._specialized_algorithms"
+    ),
+    "AA_KNN": "._algorithm_adaptation",
+    "AA_BP": "._bp",
+    **dict.fromkeys(
+        [
+            "_PT",
+            "PT_Bayes",
+            "PT_SVM",
+        ],
+        "._problem_transformation"
+    ),
+    # -------------------- 2015 --------------------
+    "LDSVR": "._problem_transformation",
+    # -------------------- 2013 --------------------
+    "CPNN": "._cpnn",
+}
+
+_LE_MODULE_MAP = {
+    # -------------------- 2023 --------------------
+    **dict.fromkeys(
+        [
+            "LIBLE",
+            "ConLE",
+        ],
+        "._label_enhancement"
+    ),
+    # -------------------- 2020 --------------------
+    "LEVI": "._label_enhancement",
+    # -------------------- 2019 --------------------
+    **dict.fromkeys(
+        [
+            "GLLE",
+            "ML",
+            "LP",
+            "KM",
+            "FCM",
+        ],
+        "._label_enhancement"
+    ),
+}
+
+_INCOMLDL_MODULE_MAP = {
+    # -------------------- 2024 --------------------
+    "WInLDL": "._incomplete",
+    # -------------------- 2017 --------------------
+    "IncomLDL": "._incomplete",
+}
+
+_LDL4C_MODULE_MAP = {
+    # -------------------- 2021 --------------------
+    **dict.fromkeys(
+        [
+            "LDLM",
+            "LDL_HR",
+            "LDL4C",
+        ],
+        "._classifier"
+    ),
+}
+
+_SSG_LDL_MODULE_MAP = {
+    # -------------------- 2021 --------------------
+    "SSG_LDL": "._ssg_ldl"
+}
+
+_LDL_DA_MODULE_MAP = {
+    # -------------------- 2025 --------------------
+    "LDL_DA": "._ldl_da"
+}
+
+_GLD_MODULE_MAP = {
+    # -------------------- 2026 --------------------
+    "GLD_SVR": "._problem_transformation",
+    "GLD_KNN": "._algorithm_adaptation",  
+    "GLD_BFGS": "._specialized_algorithms",
+}
+
+_MODULE_MAP = {
+    **_LDL_MODULE_MAP,
+    **_LE_MODULE_MAP,
+    **_INCOMLDL_MODULE_MAP,
+    **_LDL4C_MODULE_MAP,
+    **_SSG_LDL_MODULE_MAP,
+    **_LDL_DA_MODULE_MAP,
+    **_GLD_MODULE_MAP,
+}
+
+
+_ldl__ = list(_LDL_MODULE_MAP)
+_le__ = list(_LE_MODULE_MAP)
+_incomldl__ = list(_INCOMLDL_MODULE_MAP)
+_ldl4c__ = list(_LDL4C_MODULE_MAP)
+_ssg_ldl__ = list(_SSG_LDL_MODULE_MAP)
+_ldl_da__ = list(_LDL_DA_MODULE_MAP)
+_gld__ = list(_GLD_MODULE_MAP)
+
+
+__all__ = _ldl__ + _le__ + _incomldl__ + _ldl4c__ + _ssg_ldl__ + _ldl_da__ + _gld__
+
+
+from .base._lazy import lazy_module
+__getattr__ = lazy_module(_MODULE_MAP, __name__)