Devinder Kaur

Anuj Maharjan, Devinder Kaur, Richard Molyet

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by grounding their responses in external knowledge, but conventional pipelines rely on static, single-step retrieval that limits performance on complex queries. This paper presents an Agent-Orchestrated Adaptive RAG framework that introduces dynamic query decomposition, iterative retrieval, and a bounded self-reflective evaluation loop. We evaluate the system across two complementary datasets: a domain-specific DevOps knowledge base and the multi-hop reasoning benchmark MuSiQue. Using metrics that include overall score, citation accuracy, mean reciprocal rank, and topic coverage, we find that query decomposition yields consistent gains in the structured domain (overall score $+0.04$, MRR $+0.17$ on DevOps) but degrades ranking precision on the multi-hop benchmark, while the reflection mechanism improves citation accuracy at a substantial latency cost. These contrasting results show that agentic enhancements are not universally beneficial and must be applied selectively according to query and domain characteristics. Our findings argue for adaptive, cost-aware orchestration rather than uniformly aggressive reasoning pipelines.

Devinder Kaur, Shama Naz Islam, Md. Apel Mahmud

In this paper, we propose an improved Bayesian bidirectional long-short term memory (BiLSTM) neural networks for multi-step ahead (MSA) solar generation forecasting. The proposed technique applies alpha-beta divergence for a more appropriate consideration of outliers in the solar generation data and resulting variability of the weight parameter distribution in the neural network. The proposed method is examined on highly granular solar generation data from Ausgrid using probabilistic evaluation metrics such as Pinball loss and Winkler score. Moreover, a comparative analysis between MSA and the single-step ahead (SSA) forecasting is provided to test the effectiveness of the proposed method on variable forecasting horizons. The numerical results clearly demonstrate that the proposed Bayesian BiLSTM with alpha-beta divergence outperforms standard Bayesian BiLSTM and other benchmark methods for MSA forecasting in terms of error performance.

Devinder Kaur, Shama Naz Islam, Md. Apel Mahmud et al.

The advancement of distributed generation technologies in modern power systems has led to a widespread integration of renewable power generation at customer side. However, the intermittent nature of renewable energy poses new challenges to the network operational planning with underlying uncertainties. This paper proposes a novel Bayesian probabilistic technique for forecasting renewable solar generation by addressing data and model uncertainties by integrating bidirectional long short-term memory (BiLSTM) neural networks while compressing the weight parameters using variational autoencoder (VAE). Existing Bayesian deep learning methods suffer from high computational complexities as they require to draw a large number of samples from weight parameters expressed in the form of probability distributions. The proposed method can deal with uncertainty present in model and data in a more computationally efficient manner by reducing the dimensionality of model parameters. The proposed method is evaluated using quantile loss, reconstruction error, and deterministic forecasting evaluation metrics such as root-mean square error. It is inferred from the numerical results that VAE-Bayesian BiLSTM outperforms other probabilistic and deterministic deep learning methods for solar power forecasting in terms of accuracy and computational efficiency for different sizes of the dataset.

Devinder Kaur, Shama Naz Islam, Md. Apel Mahmud et al.

Energy forecasting has a vital role to play in smart grid (SG) systems involving various applications such as demand-side management, load shedding, and optimum dispatch. Managing efficient forecasting while ensuring the least possible prediction error is one of the main challenges posed in the grid today, considering the uncertainty and granularity in SG data. This paper presents a comprehensive and application-oriented review of state-of-the-art forecasting methods for SG systems along with recent developments in probabilistic deep learning (PDL) considering different models and architectures. Traditional point forecasting methods including statistical, machine learning (ML), and deep learning (DL) are extensively investigated in terms of their applicability to energy forecasting. In addition, the significance of hybrid and data pre-processing techniques to support forecasting performance is also studied. A comparative case study using the Victorian electricity consumption and American electric power (AEP) datasets is conducted to analyze the performance of point and probabilistic forecasting methods. The analysis demonstrates higher accuracy of the long-short term memory (LSTM) models with appropriate hyper-parameter tuning among point forecasting methods especially when sample sizes are larger and involve nonlinear patterns with long sequences. Furthermore, Bayesian bidirectional LSTM (BLSTM) as a probabilistic method exhibit the highest accuracy in terms of least pinball score and root mean square error (RMSE).

Dominic Wilson, Devinder Kaur

The Santa Fe Ant model problem has been extensively used to investigate, test and evaluate Evolutionary Computing systems and methods over the past two decades. There is however no literature on its program structures that are systematically used for fitness improvement, the geometries of those structures and their dynamics during optimization. This paper analyzes the Santa Fe Ant Problem using a new phenotypic schema and landscape analysis based on executed instruction sequences. For the first time we detail systematic structural features that give high fitness and the evolutionary dynamics of such structures. The new schema avoids variances due to introns. We develop a phenotypic variation method that tests the new understanding of the landscape. We also develop a modified function set that tests newly identified synchronization constraints. We obtain favorable computational efforts compared to those in the literature, on testing the new variation and function set on both the Santa Fe Trail, and the more computationally demanding Los Altos Trail. Our findings suggest that for the Santa Fe Ant problem, a perspective of program assembly from repetition of highly fit responses to trail conditions leads to better analysis and performance.