This paper introduces the Dynamic External Memory LSTM (DEM-LSTM), a novel deep neural architecture designed to address the hidden state information bottleneck and temporal context decay inherent to standard LSTMs in financial time-series forecasting. By decoupling sequence processing from persistent state storage via an addressable external memory matrix ($M_t$), DEM-LSTM dynamically reads, erases, and updates market context across long sequences without corrupting internal hidden representations. Evaluated across four distinct asset classes—Foreign Exchange (EUR/USD), Commodities (XAU/USD and USOIL), and Cryptocurrencies (BTC/USD)—DEM-LSTM consistently outperforms standard LSTM baselines across all metrics, achieving up to a 41.4% reduction in RMSE on Gold spot prices while maintaining superior stability across high-volatility market regimes.
Consumers facing home-renovation quotes operate in a classic credence-goods market: they cannot readily verify whether a quoted price is fair, and general-purpose large language models (LLMs) are now a zero-cost place to ask. Whether LLM answers are actionable for this purpose is untested. Demand-side benchmarks exist for medical, legal, and financial advice, but not for construction costs. We present, to our knowledge, the first consumer-question benchmark for construction costs. Forty Japanese renovation-price questions were posed to frontier LLMs, with repeated-trial sets measuring output stability. A matched re-run at bare provider defaults with a current frontier model (gpt-5.5) was added to remove a settings confound present in the original configuration. Two findings are robust across models, generations, and settings: no LLM answer contained an explicit over-charge decision threshold, and repeated runs of the same question returned materially different price figures. Within-answer price spans are also wide, with a median of 10x under bare defaults. A deterministic structured engine over an open cost database is included as an existence proof that a citable reference layer is constructible. Its consistency is a design property and its accuracy is not validated here; validating it against completed real-world quotations is the next study. All questions, raw outputs, harness, and scoring code are public.
This paper empirically compares four architectures for financial time-series forecasting: LSTM, CNN, the original Regularized Self Attention Regression (RSAR) model, and a multiobjective optimized RSAR variant, denoted MO-RSAR, obtained using the Non dominated Sorting Genetic Algorithm II (NSGA-II). The models are evaluated on six datasets covering Forex, equity index and cryptocurrency markets, for short and long horizons. All models share a common preprocessing pipeline and evaluation framework and are assessed using standard error metrics, with emphasis on Mean Absolute Percentage Error (MAPE). MORSAR yields the lowest average prediction error across all datasets and provides significant gains for longer, more volatile horizons, while simpler architectures remain competitive for short-term forecasts. The key methodological contribution is the first empirical integration of the RSAR architecture with NSGA-IIbased multi-objective hyperparameter optimization for financial time-series forecasting. The proposed framework treats RSAR configuration as a bi-objective search over accuracy and generalization (via the train-validation gap), and evaluates the resulting model under a unified protocol across heterogeneous markets and horizons.
Akhter Javed, Huma Gul, Ali Husnain, Rahmat Said · 5 authors
Background: In this study, the increased complexity of today supply chains and explain why conventional forecasting and inventory management techniques are inadequate in today's dynamic and uncertain market conditions. As globalization and data increase, AI has become a gamechanger in delivering better demand forecasting and inventory management, in turn driving a better operation and cost savings. Objectives: This study seeks to assess the performance of AI-based demand forecasting models combined with inventory optimization methods on improving the overall performance of the supply chain. Methods: A quantitative, data-driven methodology was employed, and secondary data were used, including historical demand, inventory levels, and other external data that included seasonality and economic indicators. Demand forecasting models: Advanced machine learning and deep learning models such as Long Short-Term Memory (LSTM), Random Forest and Gradient Boosting were used for demand forecasting. The results of the forecasts were fed into an inventory optimization system using reinforcement learning for dynamic decision-making. Standard deviations like Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) were used to measure the model's performance along with cost-performance analysis. Results: The accuracy of the prediction is significantly higher in AI-based models, especially the LSTM model, than the traditional models, which decreases the errors of the prediction and enhances its responsiveness. AI-powered inventory optimization resulted in significant savings on inventory holding and shortage/cost of order, and improved service levels and inventory stockout rates. The use of external data had yet further improved predictive performance. Conclusion: AI-powered demand forecasting and inventory optimization offer a solid solution to improve the efficiency of the supply chain, make intelligent decisions and minimize operational costs. References Ahn, H. I., Song, Y. C., Olivar, S., Mehta, H., & Tewari, N. (2024). GNN-based probabilistic supply and inventory predictions in supply chain networks. arXiv. Albayrak Ünal, Ö., Erkayman, B., & Usanmaz, B. (2023). Applications of artificial intelligence in inventory management: A systematic review of the literature. Archives of Computational Methods in Engineering. Advance online publication. https://doi.org/10.1007/s11831-023-09977-2 Ayub, M. I., Gharami, A. K., Nitu, F. N., Uddin, M. N., Islam, M. I., Nijhum, A. M., … Yezdani, S. (2025). AI-driven demand forecasting for multi-echelon supply chains: Enhancing forecasting accuracy and operational efficiency through machine learning and deep learning techniques. Emerging Frontiers Library for The American Journal of Management and Economics Innovations, 7(7), 74–85. Cannas, V. G., Ciano, M. P., Saltalamacchia, M., & Secchi, R. (2024). Artificial intelligence in supply chain and operations management: A multiple case study research. International Journal of Production Research. Advance online publication. https://doi.org/10.1080/00207543.2024.2330633 Choi, T. M. (2022). Supply chain analytics and AI-driven forecasting. Annals of Operations Research. https://doi.org/10.1007/s10479-022-04652-6 Dolgui, A., Ivanov, D., & Sokolov, B. (2022). Reconfigurable supply chain systems. International Journal of Production Research, 60(2), 413–440. https://doi.org/10.1080/00207543.2021.1897179 Douaioui, K., Oucheikh, R., Benmoussa, O., & Mabrouki, C. (2024). Machine learning and deep learning models for demand forecasting in supply chain management: A critical review. Applied System Innovation, 7(2), 40. https://doi.org/10.3390/asi7020040 Fatima, A., & Salam, M. A. (2026). A data-driven predictive framework for inventory optimization using context-augmented machine learning models. arXiv. Ghodake, S. P., Malkar, V. R., Santosh, K., Jabasheela, L., Abdufattokhov, S., & Gopi, A. (2024). Enhancing supply chain management efficiency: A data-driven approach using predictive analytics and machine learning algorithms. International Journal of Advanced Computer Science and Applications, 15(4). Islam, M. K., Ahmed, H., Al Bashar, M., & Taher, M. A. (2024). Role of artificial intelligence and machine learning in optimizing inventory management across global industrial manufacturing and supply chain: A multi-country review. International Journal of Management Information Systems and Data Science, 1(2), 1–14. Ivanov, D., & Dolgui, A. (2021). A digital supply chain twin for managing the disruption risks and resilience in the era of Industry 4.0. International Journal of Production Research, 59(18), 5633–5645. https://doi.org/10.1080/00207543.2020.1768450 Jin, Z. L., Maasoumy, M., Liu, Y., Zheng, Z., & Ren, Z. (2025). Stochastic optimization of inventory at large-scale supply chains. arXiv. Judijanto, L., Riandari, F., & Marsoit, P. T. (2024). Leveraging AI for optimization in supply chain decision support. Jurnal Teknik Informatika. Kache, F., & Seuring, S. (2022). Challenges and opportunities of digital information at the intersection of big data analytics and supply chain management. International Journal of Operations & Production Management, 42(1), 1–30. https://doi.org/10.1108/IJOPM-02-2021-0129 Kagalwala, H., Radhakrishnan, G. V., Mohammed, I. A., Kothinti, R. R., & Kulkarni, N. (2025). Predictive analytics in supply chain management: The role of AI and machine learning in demand forecasting. Advances in Consumer Research, 2, 142–149. Kamble, S. S., Gunasekaran, A., & Sharma, R. (2023). Modeling blockchain-enabled traceability in supply chains. International Journal of Information Management, 68, 102509. https://doi.org/10.1016/j.ijinfomgt.2022.102509 Kaul, D., & Khurana, R. (2022). AI-driven optimization models for e-commerce supply chain operations: Demand prediction, inventory management, and delivery time reduction with cost efficiency considerations. International Journal of Social Analytics, 7(12), 59–77. https://doi.org/10.4018/IJSA.315876 Liu, R., & Vakharia, V. (2024). Optimizing supply chain management using hybrid AI models. Journal of Organizational and End User Computing, 36(2), 1–18. https://doi.org/10.4018/JOEUC.347356 Min, H. (2022). Artificial intelligence in supply chain management: Theory and applications. International Journal of Logistics Research and Applications, 25(3), 289–303. https://doi.org/10.1080/13675567.2020.1849508 Mitta, N. R. (2023). AI-driven optimization of supply chain networks in manufacturing: Utilizing machine learning for demand forecasting, inventory management, and logistics efficiency. Los Angeles Journal of Intelligent Systems and Pattern Recognition, 3, 404–446. Nweje, U., & Taiwo, M. (2025). Leveraging artificial intelligence for predictive supply chain management: Focus on how AI-driven tools are revolutionizing demand forecasting and inventory optimization. International Journal of Science and Research Archive, 14(1), 230–250. Pasupuleti, V., Thuraka, B., Kodete, C. S., & Malisetty, S. (2024). Enhancing supply chain agility and sustainability through machine learning: Optimization techniques for logistics and inventory management. Logistics, 8(3), 73. https://doi.org/10.3390/logistics8030073 Patil, D. (2024). Artificial intelligence-driven supply chain optimization: Enhancing demand forecasting and cost reduction (SSRN Working Paper No. 5057408). SSRN. https://doi.org/10.2139/ssrn.5057408 Queiroz, M. M., & Telles, R. (2023). Big data analytics in supply chain management: A review. Transportation Research Part E: Logistics and Transportation Review, 170, 102987. https://doi.org/10.1016/j.tre.2022.102987 Sajja, G. S., Addula, S. R., Meesala, M. K., & Ravipati, P. (2025). Optimizing inventory management through AI-driven demand forecasting for improved supply chain responsiveness and accuracy. In AIP Conference Proceedings (Vol. 3306, No. 1, Article 050003). AIP Publishing. Shahnawaz, M., & Safder, A. (2025). Stochastic learning-optimization model for resilient supply chains. arXiv. Shen, L., & Zang, Z. (2024). Enterprise supply chain network optimization algorithm based on blockchain-distributed technology. Information Discovery and Delivery. Advance online publication. Sodhi, M. S., & Tang, C. S. (2021). Supply chain management for extreme conditions. MIT Sloan Management Review, 62(2), 1–8. Tang, W. (2024). Improvement of inventory management and demand forecasting by big data analytics in supply chain. Applied Mathematics and Nonlinear Sciences, 9(1). Verma, P. (2024). Transforming supply chains through AI: Demand forecasting, inventory management, and dynamic optimization. Integrated Journal of Science and Technology, 1(3). Waller, M. A., & Fawcett, S. E. (2021). Data science, predictive analytics, and big data: A revolution that will transform supply chain design and management. Journal of Business Logistics, 34(2), 77–84. https://doi.org/10.1111/jbl.12010