Neural Architecture Search for Image Restoration
We propose a NAS approach that is multi-objective and considers goals of different nature (including performance, robustness, etc.). Our NAS method will be hybrid, combining global search based on evolution with local search based on domain knowledge.
PARTICIPANTS
Raúl Monroy Borja, Víctor Adrián Sosa Hernández
RESULTS
Articles, research proposals, short stays, conference presentations
FUNDING
CONAHCYT CF-803-2022 I
Towards Smart & Sustainable Engineering Supply Chains
The project is a collaborative initiative between the University of Leeds and Tecnológico de Monterrey, aiming to develop long-term research focused on the design of sustainable engineering supply chains. By considering both product design and manufacturing processes, the project seeks to improve supply chain performance in alignment with the United Nations Sustainable Development Goals (SDGs), particularly regarding emissions and the availability of raw materials.
The project will include workshops to develop research roadmaps and joint high-priority proposals, with the goal of strengthening engineering communities to drive sustainable development in global supply chains.
PARTICIPANTS
Rafael Batres, Eduardo Bastida, Dan Trowsdale, Omar Huerta, Jonathan Busch, Chee Yew Wong, Francisco Tapia Lara, Carlos Alberto González Almaguer
RESULTS
Artículos, estancias de investigación, reuniones de colaboración
FUNDING
The International Strategy Fund (ISF), University of Leeds
Advanced AI for Mental Disorders Recognition
This project will use an unprecedented dataset, created with the support of more than 40 users (Mexican and Canadian), to obtain attributes from three different sources: behavioral features extracted from sensors in fitness trackers of different brands, speech samples, and text analysis.
With this dataset, advanced artificial intelligence techniques will be applied to detect depression and other mental disorders by analyzing conversational audio, biomarkers, and social media posts. Machine learning and deep learning models best suited to the new dataset will be developed to determine a person’s psychological state with high accuracy.
In addition, the dataset will be made available to the scientific community for future studies on the subject.
At the core of the research, the development of a platform based on wearables and embedded AI models is envisioned, capable of alerting users in real time about psychophysiological symptoms related to mood disorders. Thanks to this preliminary assessment, individuals can be promptly referred to the appropriate psychological service, where trained professionals can assist them in preventing mental health symptoms from negatively impacting their life and environment.
PARTICIPANTS
Tec de Monterrey Luis A. Trejo, Miryam Villa Pérez, INAOE: Luis Villaseñor; Centro GEO: Daniela Moctezuma
ProAmbitIon: Online Process Conformance Checking with Ambiguities Driven by the Internet of Things
The ProAmbitIon project considers the use of the Internet of Things (IoT) as an enabler to close the gap between physical world process execution and its digital representation. When considering human-centered processes, process knowledge and descriptions are often provided in unstructured informal documents that allow multiple valid (ambiguous) interpretations and executions. To tackle these problems, we leverage techniques from Process Mining, Internet of Things, Generative AI, and Computer Vision to monitor processes in cyber-physical systems (e.g., healthcare and manufacturing). The ultimate goal is to pinpoint sudden deviations and report them to stakeholders through intuitive, easy-to-interpret feedback.
PARTICIPANTS
Luciano García Bañuelos, Enrique García Ceja, César Torres Huitzil (Professors)
Mauricio Jacobo González González, Astrid Monserrat Rivera González (PhD students)
Raúl Jiménez Cruz, Pedro Aarón Hernández Ávalos, Luis Armando Rodríguez (Post-docs)
ACADEMIC PARTNERS
Barbara Weber, Marco Franceschetti, Ronny Seiger — University of St. Gallen, Switzerland
Abel Armas — The University of Melbourne, Australia
RESULTS
Publications in international conferences and journals (BPM, CAiSE, CoopIS, ER)
Datasets published on Zenodo
arXiv preprints
Papers under submission to journals
SELECTED PUBLICATIONS
Franceschetti et al. (2026). On-the-fly Event Disambiguation via Alignments. BPMDS, Springer. (Runner-up Best Paper Award)
Jiménez Cruz et al. (2026). Multimodal Interpretable Feedback for Phlebotomy Training. BPMDS, Springer.
Franceschetti et al. (2026). Multimodal Process Monitoring with On-Demand Disambiguation. BPM, Springer.
García-Bañuelos et al. (2025). A semi-automated approach to detecting process-level activities from sensor data. Procedia Computer Science.
Rivera-Partida et al. (2024). All optimal K-bounded alignments using the FM-index. CoopIS, Springer.
Franceschetti et al. (2023). ProAmbitIon: Online process conformance checking with ambiguities driven by the IoT. CAiSE.
Franceschetti et al. (2023). A characterisation of ambiguity in BPM. ER, Springer.
DATASETS
Franceschetti et al. (2026). On-the-fly Event Disambiguation via Alignments. doi:10.5281/zenodo.17276498
Jiménez Cruz et al. (2025). Labeled Images of a Simulated Phlebotomy Procedure. doi:10.5281/zenodo.16924785
Seiger et al. (2024). Dataset from a Smart Factory. doi:10.5281/zenodo.14441996
Franceschetti et al. (2023). A characterisation of ambiguity in BPM. doi:10.5281/zenodo.7944319
ONGOING WORK
Jiménez Cruz et al. (2026). A Labeled Dataset of Simulated Phlebotomy Procedures for Medical AI. Submitted to Data in Brief.
González et al. (2026). Discovery of Process Activities from IoT Data using Error-tolerant Automata. To be submitted to Future Generation Computer Systems.
Hernández-Ávalos & García-Bañuelos (2026). Pragmos: A Process Agentic Modeling System. arXiv: https://doi.org/10.48550/arXiv.2604.27311
Rodríguez-Flores et al. (2026). Secure Conformance Checking using Homomorphic Encryption. arXiv: https://doi.org/10.48550/arXiv.2604.25190
Rodríguez-Flores et al. (2026). A Privacy-Preserving Approach to Conformance Checking. arXiv: https://doi.org/10.48550/arXiv.2605.00283