fix citations

Author: kameronton

paper.bib

@@ -1,9 +1,15 @@
 @software{jax2018github,
-  author = {James Bradbury and Roy Frostig and Peter Hawkins and Matthew James Johnson and Chris Leary and Dougal Maclaurin and George Necula and Adam Paszke and Jake Vander{P}las and Skye Wanderman-{M}ilne and Qiao Zhang},
+
+  author = {James Bradbury and Roy Frostig and Peter Hawkins and Matthew James Johnson and Yash Katariya and Chris Leary and Dougal Maclaurin and George Necula and Adam Paszke and Jake Vander{P}las and Skye Wanderman-{M}ilne and Qiao Zhang},
+
   title = {{JAX}: composable transformations of {P}ython+{N}um{P}y programs},
+
   url = {http://github.com/jax-ml/jax},
+
   version = {0.3.13},
-  year = {2018}
+
+  year = {2018},
+
 }
 
 @software{flax2020github,
@@ -86,14 +92,30 @@ @article{silva2020optimization
   organization = {PMLR}
 }
 
+@inproceedings{panda2024vanilla,
+  author       = {Subrat Prasad Panda and
+                  Blaise Genest and
+                  Arvind Easwaran and
+                  Ponnuthurai Nagaratnam Suganthan},
+  title        = {Vanilla Gradient Descent for Oblique Decision Trees},
+  booktitle    = {{ECAI} 2024 - 27th European Conference on Artificial Intelligence},
+  series       = {Frontiers in Artificial Intelligence and Applications},
+  volume       = {392},
+  pages        = {1140--1147},
+  year         = {2024},
+  url          = {https://doi.org/10.3233/FAIA240607},
+  doi          = {10.3233/FAIA240607},
+}
+
 @article{topin2021iterative,
   title = {Iterative Bounding {MDP}s: Learning Interpretable Policies via Non-Interpretable Methods},
   author = {Nicholay Topin and Stephanie Milani and Fei Fang and Manuela Veloso},
   journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
   volume = {35},
   number = {11},
   pages = {9923--9931},
-  year = {2021}
+  year = {2021},
+  doi = {10.1609/aaai.v35i11.17192}
 }
 
 @article{landajuela2021discovering,

paper.md

@@ -24,7 +24,7 @@ bibliography: paper.bib
 # Summary
 
 BordAX is a Python framework for reinforcement learning (RL) built on JAX [@jax2018github].
-It provides a modular, fully JIT-compilable training pipeline that supports multiple policy representations, including standard multilayer perceptrons (MLPs), differentiable decision trees (DTSemNet), and boolean function networks (HyperBool).
+It provides a modular, fully JIT-compilable training pipeline that supports multiple policy representations, including standard multilayer perceptrons (MLPs), differentiable decision trees (DTSemNet, [@panda2024vanilla]), and boolean function networks (HyperBool).
 BordAX currently implements Proximal Policy Optimization (PPO) [@schulman2017proximal] and Deep Q-Networks (DQN) [@mnih2015human], and is designed so that additional algorithms can be composed from interchangeable collector, batch-builder, and updater components.
 The framework supports both Gymnax [@gymnax2022github] environments, which enable full JIT compilation of the training loop via `jax.lax.scan`, and standard Gymnasium [@towers2026gymnasium] environments.
 On a CartPole-v1 benchmark with identical hyperparameters, BordAX with Gymnax achieves approximately 2.8 times higher throughput than Stable-Baselines3 [@raffin2021stable].