Sitemap
A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.
Pages
Posts
How you play games, is how you do everything
Published:
This train of thought was inspired by the YouTuber doozy speaks, with his video titled - How you play games, is how you do everything. After listening to his introspective video essay on his play style, I decided to roughly jot down my own play style. Patterns of behaviour, whether in small or large consequential circumstances, define the principles that govern your life. Who would have thunk?
Life/Luck Abroad
Published:
A year has passed since I’ve moved to Ireland. In this article, I share the life/luck I’ve had so far.
An Introduction to Algorithmic Bias
Published:
A blog post that I wrote for my AI and Ethics module during my Master’s.
Deep Learning on the Apple Silicon
Published:
This article will walk you through the setup procedure, training and logging on an Apple Silicon device.
My Journey with Artificial Intelligence
Published:
I summarise my journey with AI so far.
My Experience as the Captain of the Football Team
Published:
In this blog post, I write about my experience being the captain of the football team.
Getting into Habits
Published:
I write about how I plan to get into habits.
publications
FGTD: Face Generation from Textual Description
Published in Inventive Communication and Computational Technologies: Proceedings of ICICCT 2021, 2022
The majority of current text-to-image generation tasks are limited to creating images like flowers (Oxford 102 Flower), birds (CUB-200–2011), and common objects (COCO) from captions. The existing face datasets such as Labeled Faces in the Wild and MegaFace lack description while datasets like CelebA have attributes associated but do not provide feature descriptions. Thus, in this paper, we build upon an existing algorithm to create captions with the attributes provided in the CelebA dataset, which can not only generate one caption, but it can also be extended to generate N captions per image. We utilize sentence BERT to encode these descriptions into sentence embeddings. We then perform a comparative study of three models-DCGAN, SAGAN, and DFGAN, by using these sentence embeddings along with a latent noise as the inputs to the different architectures. Finally, we calculate the Inception Scores and the FID values to compare the output images across different architectures.
Recommended citation: Deorukhkar, K., Kadamala, K., & Menezes, E. (2022). FGTD: face generation from textual description. In Inventive Communication and Computational Technologies: Proceedings of ICICCT 2021 (pp. 547-562). Singapore: Springer Nature Singapore.
Download Paper
An Insight into NeuroEvolution and Genetic Algorithms for Text Classification
Published in International Conference on Knowledge Based and Intelligent Information and Engineering Sytems (KES), 2023
Natural Language Processing (NLP) systems have, over the past decade, shifted from using rule-based techniques to using machine learning-based algorithms. This has led to the development of different architectures and models for different tasks. Some of these architectures include models like the transformers, the CNN and the RNN, which have now become ubiquitous in NLP. However, designing these neural network architectures usually requires in-depth analysis and knowledge of multiple domain areas involved with the problem at hand. In our work, we evaluate an alternative solution to this problem in the domain of text classification. Here, we suggest using the Genetic Algorithm with gradient descent (GAGD) and NeuroEvolution of Augmenting Topologies (NEAT) to search for an optimal neural architecture for the Reuters-21578 and 20 Newsgroups datasets. We evaluate and compare the results of the two algorithms against the current state-of-the-art architectures and provide insight into their performance.
Recommended citation: Kadamala, K., & Griffith, J. (2023). An Insight into NeuroEvolution and Genetic Algorithms for Text Classification. Procedia Computer Science, 225, 1379-1387.
Download Paper
Enhancing HVAC control systems through transfer learning with deep reinforcement learning agents
Published in Smart Energy, 2024
Traditionally, building control systems for heating, ventilation, and air conditioning (HVAC) relied on rule-based scheduler systems. Deep reinforcement learning techniques have the ability to learn optimal control policies from data without the need for explicit programming or domain-specific knowledge. However, these data-driven methods require considerable time and data to learn effective policies without prior knowledge. Performing transfer learning using pre-trained models avoids the need to learn the underlying data from scratch, thus saving time and resources. In this work, we evaluate reinforcement learning as a method of pretraining and fine-tuning neural networks for HVAC control. First, we train an RL agent in a building simulation environment to obtain a foundation model. We then fine-tune this model on two separate simulation environments such that one environment simulates the same building under different weather conditions while the other environment simulates a different building under the same weather conditions. We perform these experiments with two different reward functions to evaluate their effect on transfer learning. The results indicate that transfer learning agents outperform the rule-based controller and show improvements in the range of 1% to 4% when compared to agents trained from scratch.
Recommended citation: Kadamala, K., Chambers, D., & Barrett, E. (2024). Enhancing HVAC control systems through transfer learning with deep reinforcement learning agents. Smart Energy, 13, 100131.
Download Paper
Enhancing HVAC Control Efficiency: A Hybrid Approach Using Imitation and Reinforcement Learning
Published in Joint European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD), 2024
This paper explores the application of imitation learning (IL) and reinforcement learning (RL) in HVAC control. IL learns to perform tasks by imitating a demonstrator, utilising a dataset of demonstrations. However, the performance of IL is highly dependent on the quality of the expert demonstration data. On the other hand, RL can adapt control policies based on different objectives, but for larger problems, it can be sample inefficient, requiring significant time and resources for training. To overcome the limitations of both RL and IL, we propose a combined methodology where IL is used for pre-training and RL for fine-tuning. We introduce a fine-tuning methodology to HVAC control inspired by a robot navigation task. Using the 5-Zone residential building environment provided by Sinergym, we collect state-action pairs from interactions with the environment using a rule-based policy to create a dataset of expert demonstrations. Our experiments show that this combined methodology improves the efficiency and performance of the RL agent by 1% to 11.35% compared to existing literature. This study contributes to the ongoing discourse on how imitation learning can enhance the performance of reinforcement learning in building control systems.
Recommended citation: Kadamala, K., Chambers, D., & Barrett, E. (2024, August). Enhancing HVAC control efficiency: a hybrid approach using Imitation and reinforcement learning. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases (pp. 256-270). Cham: Springer Nature Switzerland.
Download Paper
Transfer Learning with TD3 for Adaptive HVAC Control in Diverse Building Environments
Published in Highlights in Practical Applications of Agents, Multi-Agent Systems, and Digital Twins: The PAAMS Collection, 2025
This paper demonstrates that applying the TD3 algorithm with transfer learning to heterogeneous HVAC control scenarios can perform competitively against rule-based and scratch-trained agents, showcasing its potential for adaptive, cross-building HVAC optimisation.
Recommended citation: Kadamala, K., Chambers, D., & Barrett, E. (2024, June). Transfer Learning with TD3 for Adaptive HVAC Control in Diverse Building Environments. In International Conference on Practical Applications of Agents and Multi-Agent Systems (pp. 256-267). Cham: Springer Nature Switzerland.
Download Paper
Improving HVAC Control with Transfer Learning: Using Padding Techniques for Cross-Building Pre-training and Fine-tuning
Published in Energy and AI, 2025
Recent advancements have shown that control strategies using Deep Reinforcement Learning (DRL) can significantly improve the management of HVAC control and energy systems in buildings, leading to significant energy savings and better comfort. Unlike conventional rule-based controllers, they demand considerable time and data to develop effective policies. Transfer learning using pre-trained models can help address this issue. In this work, we use imitation learning (IL) as a method of pre-training and reinforcement learning (RL) for fine-tuning. However, HVAC systems can vary depending on the location, building size, structure, construction materials and weather conditions. The diversity in HVAC control systems across different buildings complicates the use of IL and RL. Neural network weights trained on the source building cannot be directly transferred to the target building because of differences in input features and the number of control equipment. To overcome this problem, we propose a novel padding method to ensure that both the source and target buildings share the same state space dimensionality. Thus, the trained neural network weights are transferable, and only the output layer must be adjusted to fit the dimensionality of the target action space. Additionally, we evaluate the performance of an existing padding technique for comparison. Our experiments show that the novel padding technique outperforms zero padding by 1.37% and training from scratch by 4.59% on average.
Recommended citation: Kadamala, K., Chambers, D., & Barrett, E. (2025). Improving HVAC control with transfer learning: Using padding techniques for cross-building pre-training and fine-tuning. Energy and AI, 21, 100531.
Download Paper