The convolutional neural network (CNN) is commonly used in visual recognitions and classifications. However, CNN can also be applied as a forecaster that can extract features from spatiotemporal data. This paper proposes a 24h ahead electricity price forecasting method, which integrates CNN with an evolutionary algorithm and utilizes spatiotemporal data. The optimal structure of the CNN network for the locational marginal price (LMP) forecasting was obtained using a genetic algorithm (GA). A gene mapping scheme was initially encoded to represent the search space and the process of selection, mutation, and crossover eliminated structures that did not satisfy the validation fitness function and then competitive individuals were generated. The evolution process uses the root mean square error (RMSE) as the validation fitness function, which is optimzed by training the created CNN network. The proposed gene mapping scheme can be used to design an optimal CNN structure once the mapping between gene binary bits and parameters/hyperparameters of CNN is given. Day-ahead LMP and demand datasets from Pennsylvania-New Jersey-Maryland (PJM) power market were used to demonstrate the evolutionary capability of the proposed method and the finding of optimal CNN structures. Each studied dataset was grouped into 4 subsets corresponding to various seasonal characteristics (different types of situations in real life). Experimental results revealed that the proposed GA-CNN always yielded a higher forecasting accuracy and lower error rates than other forecasting methods.

The Philippine stock market, just like most of its neighbors in the region, was seriously impacted by the global pandemic COVID-19. Investors remain hopeful while continuing to seek great ones in the damaged market. This paper developed a methodology for portfolio selection and optimization with the use of technical analysis, machine learning techniques, and portfolio optimization model. The combined methods of technical analysis, K-means clustering algorithm, and mean-variance portfolio optimization model will result in the development of the proposed TAKMV method. The study aims to integrate these three important analyses to identify portfolio investments. This paper uses the average annual risk and annual rate of return data for the years 2018 and 2020 to form the clusters and assessed the stocks that correspond to the investor's technical strategy such as Moving Average Convergence/Divergence (MACD) and Hybrid MACD with Arnaud Legoux Moving Average (ALMA). This paper solved the risk minimization problem on selected shares of the companies, based on the mean-variance portfolio optimization model. There are 230 and 239 companies for 2018 and 2020, respectively, listed in Philippine Stock Market, and all simulations were performed in MATLAB environment platform. Results showed that MACD strategy dominates the MACD-ALMA strategy in terms of the number of assets with a positive annual rate of return. The MACD works efficiently in the pre-COVID-19 condition while MACD-ALMA works efficiently during-COVID-19 condition, regardless of the number of assets with a positive annual rate of return. The results also show that the maximum expected portfolio return (R_P) can be achieved using the MACD and MACD-ALMA in the pre-and during-COVID-19 conditions, respectively. The MACD-ALMA shows an advantage during high-risk market conditions and can also provide maximum R_P. The performance of the TAKMV method was validated by applying its results and comparing it to the next year's historical price. The 2018 results were compared to 2019 data and the 2020 results were compared to 2021 data. For consistency, the comparison was applied to the same company per portfolio. Simulation results show that the MACD strategy is more effective compared to MACD-ALMA.

Roof Classification is an approach currently utilized by Southeast Asian countries as a tool in assisting decision makers during the design process of urban policies and development which aims to ease congestion and proper allocation of land properties. However, the Philippines is among the countries who have yet to implement this into its urbanization programs. Data augmentation and SMOTE is performed to address the potential low accuracy issue that limits the existing CNN model for roof classification. A series of experimentation to determine the ideal number of epochs and batch size were performed to acquire the optimal accuracy for the roof classification. Consequently, the model achieved a promising result in terms of its classification performance and accuracy. The CNN model exhibits strong metrics across precision, recall, and F1-score with an average accuracy of 97%.

The proliferation of Large Language Models (LLMs) is often constrained by their significant computational and memory requirements, limiting their deployment to large data centers. Small Language Models (SLMs) offer a feasible solution for on-device applications; yet their efficiency requires optimization to operate well on resource-constrained hardware. This study looks at ways to make SLMs more efficient at using computers. The effects of two primary methods were compared: post-training optimization and architectural innovation through quantitative and observational study. Using a standardized suite of benchmarks measuring accuracy, reasoning, and inference performance, a baseline is established with state-of-the-art SLMs like Phi-3 and Llama 3. Post-training techniques were evaluated, including 4-bit quantization (GPTQ) and knowledge distillation from a superior teacher model. Finally, these optimized Transformers were compared against a custom-trained, non-Transformer model based on the Mamba (State-Space Model) architecture. Results show that 4-bit quantization is the most effective compression strategy among those evaluated. It reduces peak inference memory footprint by 71%, increases throughput by 83%, and does so with minimal accuracy degradation. Within the controlled experimental space evaluated in this study, the 4-bit quantized Phi-3-mini model occupies a Pareto-optimal position in memory-normalized accuracy. Focused skill growth works best with knowledge distillation. However, new designs like Mamba offer a different trade-off by being the best at streaming jobs’ raw output. It was found that improving current Transformer-based SLMs through quantization is the best way to use them for general purposes. However, customized designs and distillation work better for specific, high-performance uses. This research offers a definitive framework and pragmatic recommendations for advancing the next generation of robust, efficient, and accessible language models.

Selecting materials for low-cost housing requires simultaneous consideration of economic constraints, environmental burdens, technical performance, durability, and implementation feasibility. This article presents a comparative literature-based synthesis of a retained 29-source corpus with a recent-literature emphasis on 2021–2026 rather than a comprehensive historical boundary. One seminal 2019 review was retained for conceptual continuity. Because original search logs, duplicate-removal records, and exclusion lists were unavailable, the corpus is neither exhaustive nor fully reproducible and the article is not presented as a systematic or scoping review. Seven analytical material categories were compared using five author-defined criteria: affordability and life-cycle cost; environmental performance; technical performance and durability; implementation feasibility; and sustainability and circularity relevance. Environmental performance was most strongly represented, whereas complete life-cycle cost, long-term durability, code acceptance, and verified implementation capacity were less consistently documented. Conventional systems retained advantages in standards, supply chains, construction familiarity, and scale. Low-carbon conventional systems offered a potentially pragmatic transition pathway, while earth-, bamboo-, recycled-, plastic-waste-, and bio-based materials showed context-dependent benefits that varied with intended function, climate, treatment, quality control, maintenance, and regulatory acceptance. Direct cross-material ranking was constrained by heterogeneous functional units, life-cycle boundaries, service-life assumptions, transport distances, and maintenance scenarios. A preliminary adaptive material selection framework is proposed using transparent 1–5 scoring, context-specific weighting, evidence-confidence flags, and non-compensatory safety checks. The framework is illustrative, nonoperational, and nonvalidated; expert elicitation, stakeholder consultation, regional case testing, sensitivity analysis, and comparison with actual housing decisions are required before project-level use.