Abstract
As ESG regulations gain relevance, namely following the entry into force of EU's Corporate Sustainability Reporting Directive, there is an increased requirement for value chain transparency. Organizations that are heavily based on data, Artificial Intelligence or Machine Learning, may significantly contribute to the ESG profiles of their upstream clients through the resources they spend on data storage and processing, model training or model serving. However, the lack of standardized mechanisms to report ESG-relevant indicators from these organizations hinders effective integration into their clients' ESG disclosures. This paper is motivated by an identified need to facilitate ESG indicator reporting by ML organizations in the value chain, and proposes an architecture to automate the monitoring and reporting process. This architecture is based on the principle of observability, and on the integration and effective use of open-source tools to monitor data processing pipelines and MLOps. By implementing an automated observability-based monitoring framework, ML organizations can provide actionable ESG data that align with regulatory requirements, while reducing reporting overhead for clients. We address how this architecture can be seamlessly integrated into existing operational workflows of ML providers and their clients, enhancing transparency, accountability, and compliance. The proposed architecture ensures accurate, real-time reporting and creates a scalable foundation for ESG aligned innovation in Machine Learning and adjacent domains. © 2025 IEEE.

Abstract
The increasing prevalence of ML in industrial environments is driven by the growing availability of userfriendly frameworks and industrial data. Manufacturing Execution Systems (MES) enabled easy data collection and utilization for decision support, namely for anomaly detection, quality control, or object detection/classification. However, models for new ML problems are often trained without regard for previous models or data, potentially wasting resources and hindering knowledge transfer. This is due to a lack of systematic methods for identifying and leveraging relevant prior knowledge. In this paper, we propose an approach designed to address this inefficiency by reusing previously trained models in new ML tasks. We reuse models based on data similarity metrics to create ensembles on-the-fly. This allows for accurate predictions on new data while minimizing the need for training from scratch. This approach has the potential to significantly reduce resource expenditure on data labeling and model training within industrial organizations.

Abstract
Explainable Artificial Intelligence (xAI) techniques are nowadays widely accepted as one of the paths towards addressing the interpretability and transparency issues of using black box models. Such techniques may allow to understand, to a certain extent, how or why a model produced a certain output, which may even help identify problems with the model or the data. As in many other domains, the use of xAI techniques in the context of manufacturing is seen as fundamental towards understanding model outputs, supporting informed decision-making, or enabling more human-centric approaches. In this paper, we specifically look at LIME, one of the most widely used approaches to xAI, and at how it needs to be adapted to the manufacturing context. Specifically, we show how the image permutations introduced by LIME might deceive the underlying model and generate poor explanations, and propose a methodology to address this issue. The specific use-case is on defect detection in the textile manufacturing industry.

Abstract
Water scarcity is currently a concerning problem and is likely to worsen in the future. To address this issue, it is essential that water used in human activities is treated before being reused or returned to nature. Wastewater is processed in wastewater treatment plants (WWTPs), which are complex structures that consume a considerable amount of resources and need to operate optimally. Many authors have proposed computational methodologies to optimize WWTPs, and each work has different approaches and characteristics, but most have in common the lack of concern with maximizing sustainability, in its broadest definition. Furthermore, even when sustainability is considered, it is typically addressed in an indirect or superficial manner, rather than being treated as a central objective. This paper provides a critical literature review of computational methodologies that, in some way, focus on improving the sustainability of WWTPs. Considering the target of the paper, this review aims to answer the following main questions: (1) What are the general objectives of the proposed works? (2) In which locations/phases of the treatment process are the proposed techniques applied? (3) What are the main methodologies and performance metrics used in the proposed techniques? The review identifies a strong focus on optimizing aeration in biological reactors, limited holistic and real-time optimization across WWTP stages, and sparse integration of sustainability metrics, especially for environmental and social impacts. Future research should prioritize the development of real-time, multi-objective optimization frameworks that encompass all WWTP stages and fully integrate economic, environmental, and social sustainability dimensions.

Abstract
The evaluation of industrial process efficiency is essential for resource optimization, enabling the identification of bottlenecks, waste, and improvement opportunities while promoting the rational use of resources and enhancing the sustainability of operations. The Multi-layer Stream Mapping (MSM) method is a tool for assessing the efficiency of complex production processes, which identifies the efficiencies and inefficiencies based on reference values. However, its limitation lies in using reference values that often fail to reflect process evolution or distinct operational regimes. This work proposes a new dynamic ecoefficiency assessment methodology, the Evolving Fuzzy Multilayer Stream Mapping System (eFuMSM), based on MSM and an evolving fuzzy system, to provide dynamic reference values, allowing more accurate eco-efficiency assessments considering the process evolution and different regions of operation. The proposed eFuMSM was applied to the primary clarifier of a wastewater treatment plant, evaluating efficiency in removing total suspended solids. Results revealed that systems previously undervalued under the traditional MSM demonstrated improved efficiency when assessed using the eFuMSM system, aligning more accurately with their operational regimes.

Abstract
This paper studies the process of cutting steel bars in a truck suspension factory with the objective of reducing its inventory costs and material losses. A mathematical model is presented that focuses on decisions for a medium-term horizon (4 periods of 2 months). This approach addresses the one-dimensional 3-level integrated lot sizing and cutting stock problem, considering demand, inventory costs and stock level limits for bars (objects-level 1), springs (items-level 2) and spring bundles (final products-level 3), as well as the acquisition of bars as a decision variable. The solution to the proposed mathematical model is reached through an optimization package, using column generation along with a method for achieving integer solutions. The results obtained with real data demonstrate that the method provides significantly better solutions than those carried out at the company, whilst using reduced computational time. Additionally, the application of tests with random data enabled the analysis of both the effect of varying parameters in the solution, which provides managerial insights, and the overall performance of the method.