Caxton Emerald S

About Me

I am a doctoral researcher in Computer Science at Pondicherry University, India, investigating the mechanistic interpretability of low-dimensional universal-approximator artificial neural networks (ANNs). My work unifies theoretical analysis with empirical results and optimisation via the JeevanRuth Neural Network Optimisation Algorithm, which reduces redundant neurons while preserving accuracy for a specific class of low dimensional ANNs. I develop and maintain the open-source python packages NNVisualiser and NeuralNetworkCoordinates, enabling fine-grained visual analysis of neuron-level transformations; corresponding model repositories are hosted on Hugging Face to support reproducibility. My research has appeared in Scopus-indexed journals and conferences. One of my research presentations at ICACS 2025 has gotten the Best Research Paper Award. Broadly, I aim to advance responsible, explainable AI by revealing how compact ANN architectures learn and generalise. Apart from my interests in Mechanistic Interpretability, I also aim to contribute towards AGI. My learning interests range from the Quantum realms of physics to Black holes of the Verse, from Cognitive Neuroscience to Evolutionary Developmental Biology.

Academic Qualifications

PhD in Computer Science (August 2019 – Present)
Pondicherry University (A Central University), Puducherry, India

Five-Year Integrated M.Sc. in Computer Science (Graduated May 2015)
Pondicherry University (A Central University), Puducherry, India
Final CGPA: 9.57

Areas of Research Interest

• Artificial Intelligence
• Mechanistic Interpretability
• Cognitive Neuroscience
• Quantum AI

Future Research Directions

• Extend mechanistic interpretability studies to high dimesnional models and LLMs
• Contribute to AGI by understanding Cognitive Neuroscience and Quantum AI

Publications

1. S. Caxton Emerald and T. Vengattaraman, “Neural network optimization via inactive neuron removal and neuron merging in single input single output architectures,”
   International Journal of Information Technology, May 2025. DOI: 10.1007/s41870-025-02545-6
2. Caxton Emerald S and V. T, “NNVisualiser: An Emerging Framework for Visualising Artificial Neural Networks,” in 2025 International Conference on Data Science, Agents & Artificial Intelligence (ICDSAAI), Chennai, India, IEEE, Mar. 2025, pp. 1–6. DOI: 10.1109/ICDSAAI65575.2025.11011573

Presented at International Conferences and Awaiting Publications

3. Caxton Emerald S, T. Vengattaraman, “Approximating the sine function with the universal approximator: An experimental study with visualisations,” International Conference on Artificial Intelligence, Communication Technologies & Smart Cities (ICACS), Mar. 01 2025, Springer.
4. Caxton Emerald S, T. Vengattaraman, “Understanding the role of neurons and layers in one-dimensional three-class classifiers,” ICACS, Mar. 01 2025, Springer — Awarded Best Research Paper.
5. Caxton Emerald S, T. Vengattaraman, “Understanding the role of Neurons and Layers in Approximating the Absolute Value Function using Artificial Neural Networks,” International Conference on AI Systems and Sustainable Technologies (IISU-ASSET), Mar. 28–29 2025, Springer.
6. Caxton Emerald S, T. Vengattaraman, “Neural Network Coordinates: A Python Package for Studying Artificial Neural Networks,” IISU-ASSET, Mar. 28–29 2025, Springer.

Other Publications

1. S. Caxton Emerald and T. Vengattaraman, “Chaotic Ant Swarm based Feature Subset Selection with Concept Drift Detection and Classification Model for Data Streaming Applications,” 2022 Second International Conference on Artificial Intelligence and Smart Energy (ICAIS), 2022, pp. 991–996. DOI: 10.1109/ICAIS53314.2022.9742928
2. S. C. Emerald and T. Vengattaraman, “Concept Drift Detection with Optimal Machine Learning Model for Data Classification,” 2022 6th International Conference on Trends in Electronics and Informatics (ICOEI), 2022, pp. 1160–1165. DOI: 10.1109/ICOEI53556.2022.9776949
3. S. C. Emerald and T. Vengattaraman, “Explainable Artificial Intelligence with Single Layer Feedforward Neural Network and Improved Crowned Porcupine Optimization Algorithm for Classification Problems,” Engineering, Technology & Applied Science Research, vol. 15, no. 2, pp. 21593–21598, Apr. 2025. DOI: 10.48084/etasr.10070
4. P. Govindharaj, N. Alungal, C. Emerald. S, and Kannan. S, "Au@CeO2 nanozyme based smart colourimetric sensor for cholesterol: A neural network powered point of care solution model", Biochemical Engineering Journal, vol. 225, p. 109908, 2026. DOI:10.1016/j.bej.2025.109908

Python Packages

• NeuralNetworkCoordinates: Precise coordinates for visualising intricate neural network transformations. – Accessed Jun. 04 2025. PyPI • GitHub
• NNVisualiser: A Python package utilising Matplotlib to visualise coordinates from NeuralNetworkCoordinates for single-input, single-output neural networks. PyPI • GitHub

Model Repository on HuggingFace Space

Simple ANNs Optimised With JeevanRuth NN Optimisation Algorithm

Research Credits

• Google Cloud Research Credits program with EDU Credit—wilsonjessica—281056349 award.

Professional Qualification

• India’s National Eligibility Test (UGC-NET June 2015) for Assistant Professorship.
• Regional Eligibility Test (TN SET July 2018) for Assistant Professorship.
• Software Engineering Training (SET) Programme by HTC Global Services on Java EE (Struts, Spring, Web Services).

Industrial Experience

Software Development Engineer — HTC Global Services, Chennai (Dec. 2016 - Mar. 2018)

• Developed real-time Java-based web applications.
• Collaborated on back-end services using Java EE technologies.

Teaching Experience

Guest Faculty — Department of Computer Science, Pondicherry University (Nov. 2015 – May 2016)

• Taught Postgraduate courses in Automata Theory and Design and Analysis of Algorithms.
• Supervised student projects and practical sessions.

Assistant Professor of Computer Science — Theivanai Ammal College for Women, Villupuram (Apr. 2018 - Apr. 2019)

• Taught courses in Cloud Computing, Android Application Development, and Programming.
• Mentored undergraduate and postgraduate students on projects and outreach activities.

Projects

• PostGrad Major Project: “A Novel Hybrid Computation Based Visual Secret Sharing Scheme with Meaningful Shares” – original research, implemented in Java.
• Grad Level Mini Project: “Latin Squares – A Simple Java Powered Gaming Application.”

Certifications

• Completed How Things Work: An Introduction to Physics (University of Virginia, Coursera).
• Completed Industrial IoT on Google Cloud Platform (Google Cloud, Coursera).
• Completed Neural Network from Scratch in TensorFlow (Coursera Project Network).

Declaration

I hereby declare that the above written particulars are true and correct to the best of my knowledge and belief.

Date: June 2025
Caxton Emerald S