myMNIST: Benchmark of PETNN, KAN, and Classical Deep Learning Models for Burmese Handwritten Digit Recognition

Ye Kyaw Thu; Thazin Myint Oo; Thepchai Supnithi

cs.CVMar 19, 2026

myMNIST: Benchmark of PETNN, KAN, and Classical Deep Learning Models for Burmese Handwritten Digit Recognition

Authors: Ye Kyaw Thu, Thazin Myint Oo, Thepchai Supnithi

Organizations: National Electronics and Computer Technology Center (NECTEC), Pathumthani, Thailand · Language Understanding Lab., Yangon, Myanmar

Abstract

We present the first systematic benchmark on a standardized iteration of the publicly available Burmese Handwritten Digit Dataset (BHDD), which we have designated as myMNIST Benchmarking. While BHDD serves as a foundational resource for Myanmar NLP/AI, it lacks a comprehensive, reproducible performance baseline across modern architectures. We evaluate eleven architectures spanning classical deep learning models (Multi-Layer Perceptron, Convolutional Neural Network, Long Short-Term Memory, Gated Recurrent Unit, Transformer), recent alternatives (FastKAN, EfficientKAN), an energy-based model (JEM), and physics-inspired PETNN variants (Sigmoid, GELU, SiLU). Using Precision, Recall, F1-Score, and Accuracy as evaluation metrics, our results show that the CNN remains a strong baseline, achieving the best overall scores (F1 = 0.9959, Accuracy = 0.9970). The PETNN (GELU) model closely follows (F1 = 0.9955, Accuracy = 0.9966), outperforming LSTM, GRU, Transformer, and KAN variants. JEM, representing energy-based modeling, performs competitively (F1 = 0.9944, Accuracy = 0.9958). KAN-based models (FastKAN, EfficientKAN) trail the top performers but provide a meaningful alternative baseline (Accuracy ~0.992). These findings (i) establish reproducible baselines for BHDD across diverse modeling paradigms, (ii) highlight PETNN's strong performance relative to classical and Transformer-based models, and (iii) quantify the gap between energy-inspired PETNNs and a true energy-based model (JEM). We release this benchmark to facilitate future research on Myanmar digit recognition and to encourage broader evaluation of emerging architectures on regional scripts.

Figures & tables

Fig. 1 : Visualization of the BHDD dataset. The figure displays five randomly selected samples for each of the ten Burmese digit classes (0-9), illustrating the variety of handwriting styles present in the dataset.

Model Type	Time per Epoch	Total Time	GPU Memory (GB)
MLP	$\sim$ 45 sec	$\sim$ 38 min	$\sim$ 1.2
CNN	$\sim$ 50 sec	$\sim$ 42 min	$\sim$ 1.5
LSTM/GRU	$\sim$ 90 sec	$\sim$ 75 min	$\sim$ 2.8
Transformer	$\sim$ 110 sec	$\sim$ 92 min	$\sim$ 3.5
JEM	$\sim$ 180 sec	$\sim$ 150 min	$\sim$ 4.0
FastKAN/EfficientKAN	$\sim$ 70 sec	$\sim$ 58 min	$\sim$ 2.0

TABLE I : Training Time and GPU Memory Utilization by Model Type (50 Epochs)

Model	Layers	Hidden Dim
MLP	3	[256, 128]
CNN	4	[32, 64]
LSTM	2	192
GRU	2	192
Transformer	2	64
JEM	5	[16, 32, 64]

TABLE II : Model architecture specifications

Model	Precision	Recall	F1-Score	Accuracy
MLP	0.9810	0.9895	0.9852	0.9907
CNN	0.9955	0.9963	0.9959	0.9970
LSTM	0.9907	0.9942	0.9924	0.9951
GRU	0.9886	0.9934	0.9910	0.9937
Transformer	0.9898	0.9944	0.9921	0.9946
JEM	0.9931	0.9957	0.9944	0.9958

TABLE III : Performance comparison of various models on the BHDD dataset.

Panel of Figure 5 — Fig. 2 : Confusion matrices for the evaluated models, highlighting common misclassification patterns.

Explore similar work

Mar 23, 2026

BHDD: A Burmese Handwritten Digit Dataset

We introduce the Burmese Handwritten Digit Dataset (BHDD), a collection of 87,561 grayscale images of handwritten Burmese digits in ten classes. Each image is 28x28 pixels, following the MNIST format. The training set has 60,000 samples split evenly across classes; the test set has 27,561 samples with class frequencies as they arose during collection. Over 150 people of different ages and backgrounds contributed samples. We analyze the dataset's class distribution, pixel statistics, and morphological variation, and identify digit pairs that are easily confused due to the round shapes of the Myanmar script. Simple baselines (an MLP, a two-layer CNN, and an improved CNN with batch normalization and augmentation) reach 99.40%, 99.75%, and 99.83% test accuracy respectively. BHDD is available under CC BY-SA 4.0 at https://github.com/baseresearch/BHDD

Swan Htet Aung, Hein Htet, Htoo Say Wah Khaing +1

Expa.AI Research Team, Yangon, Myanmar

Optical Character Recognition Image Classification

Ask PDF

Jul 15, 2026

Barnamala: Parameter-Efficient Handwritten Devanagari Recognition at Benchmark Saturation

We built a compact convolutional network (1.11 M parameters) for 46-class DHCD Devanagari recognition and reached 99.73%, the highest reported at 15.6x smaller than prior state-of-the-art. We have effectively reached the saturation point: every model tested, large teacher ensembles included, hits the same 11-error intrinsic floor. No configuration achieves a statistically clear win under exact McNemar tests with Wilson confidence intervals. Even without knowledge distillation, our student matches the nearest large-model baseline (17.32 M parameters; McNemar

p = 0.345

). Outside of DHCD, zero-shot on CMATERdb digits gives 76.6% and fine-tuning reaches 97.8%; corruption robustness is also far better than large baselines (mean corruption accuracy 75.7% vs. 38.7%). All artifacts are at https://github.com/Ampixa/barnamala.

Ashish Thapa, Samrat Karki

Ampixa Labs · Pulchowk Campus

Corruption Robustness Convolutional Neural Networks Lightweight CNNs Optical Character Recognition

Ask PDF

Feb 28, 2026

BornoViT: A Novel Efficient Vision Transformer for Bengali Handwritten Basic Characters Classification

Handwritten character classification in the Bengali script is a significant challenge due to the complexity and variability of the characters. The models commonly used for classification are often computationally expensive and data-hungry, making them unsuitable for resource-limited languages such as Bengali. In this experiment, we propose a novel, efficient, and lightweight Vision Transformer model that effectively classifies Bengali handwritten basic characters and digits, addressing several shortcomings of traditional methods. The proposed solution utilizes a deep convolutional neural network (DCNN) in a more simplified manner compared to traditional DCNN architectures, with the aim of reducing computational burden. With only 0.65 million parameters, a model size of 0.62 MB, and 0.16 GFLOPs, our model, BornoViT, is significantly lighter than current state-of-the-art models, making it more suitable for resource-limited environments, which is essential for Bengali handwritten character classification. BornoViT was evaluated on the BanglaLekha Isolated dataset, achieving an accuracy of 95.77%, and demonstrating superior efficiency compared to existing state-of-the-art approaches. Furthermore, the model was evaluated on our self-collected dataset, Bornomala, consisting of approximately 222 samples from different age groups, where it achieved an accuracy of 91.51%.

Rafi Hassan Chowdhury, Naimul Haque, Kaniz Fatiha

Department of Computer Science and Engineering, Islamic University of Technology, Gazipur 1704, Bangladesh

Optical Character Recognition Efficient ViTs Vision Transformer

Ask PDF