Keynote Talks

Sparse Matrix Algorithms for Evolving Neural Networks

Carlos Ordonez
Department of Computer Science, University of Houston, USA

Abstract

There is significant research into sparse and dense matrix computations, with high-performance computing techniques, reducing time complexity and improving parallel speedup mostly with ample main memory, considering I/O on secondary storage as a less important aspect. On the other hand, there has been important work in the database, data mining and big data communities accelerating the computation of machine learning models on large data sets. However, massive neural networks and constantly changing data sets are pushing matrix computation demands further. We first present a survey on three key problems identifying research issues: maintaining a large data set updated under frequent matrix entry insertions and deletions, sparse matrix addition/multiplication and recomputing a deep neural network when a sparse data set changes frequently. We then propose a research agenda focusing on those three major problems solved with parallel I/O efficient algorithms storing and processing matrices with coordinate tuples (like a database relational table): matrix entry insertion/deletion, matrix addition/multiplication and assembling these algorithms into state-of-the-art neural networks. We argue coordinate tuples complement and can potentially replace established main memory storage mechanisms, like dense arrays and compressed row/column formats. In summary, we believe database-inspired, parallel I/O efficient, algorithms tailored for sparse matrices can help updating, explaining and monitoring evolving neural networks on large dynamic data sets. 

Biography Carlos Ordonez studied at UNAM (top research university in Mexico), getting a B.Sc. in applied math and an M.S. in computer science. He continued PhD studies at the Georgia Institute of Technology, focusing on optimizing machine learning and data mining analysis with parallel algorithms, removing main memory limitations and enhancing their accuracy. During his PhD, Carlos joined NCR Corporation collaborating in the optimization of machine learning, statistics and cubes on the Teradata parallel DBMS, under an SMP distributed architecture. After working almost 10 years years at NCR, Carlos joined the Department of Computer Science at the University of Houston, where he currently leads the Data-Intensive Parallel Algorithms for AI group, focusing on making theory practical. During 2012 and 2013 Carlos regularly visited MIT, collaborating with Turing award winner, Michael Stonebraker, working on new-generation parallel DBMSs (columnar, array, lockfree transactions) to solve large-scale linear algebra and graph algorithms. Carlos worked as a visiting researcher with ATT Labs-Research (formerly the famous ATT Bell Labs where C, C++ and Unix were invented), where he conducted research on analyzing streams with statistical methods on massive data sets. His research projects have been funded by NSF and NIH grants.

From Data Silos to Data Mesh: A Case Study in Financial Data Architecture

Mariusz Sienkiewicz
Director, Supervisory Data Analysis Center, Polish Financial Supervision Authority, Poland

Abstract

Successful data analytics implementation requires seamless access to both data and related metadata. In many organizations, analytics challenges arise from Data Silos, which impede cross-functional access to data and knowledge sharing across the organization. This talk presents practical insights from a data architecture transformation project conducted at a large institution with over 1,400 employees and overseeing over 2,000 market entities. The organization faced significant analytical and operational challenges due to the presence of Data Silos—isolated repositories associated with specific business areas. To address these limitations, the institution initiated a transition to a Data Mesh architecture to improve data availability and enhance analytical capabilities. This talk explains the rationale behind the persistence of silos, evaluates alternative architectural models, and justifies the choice of Data Mesh based on organizational context. Key elements of the transformation include developing a data management framework, implementing a data catalog, creating a data lake to provide data input flexibility, and establishing a common analytics platform based on Data Domains. While the project is still ongoing, the talk describes the methods being implemented and shares early results, key learnings, and practical recommendations for institutions undertaking similar architectural transitions.

Biography Mariusz Sienkiewicz is a data architecture and governance expert with extensive experience in transforming large-scale data environments within highly regulated financial institutions. He currently serves as Director of the Supervisory Data Analysis Center at the Polish Financial Supervision Authority, where he leads strategic initiatives focused on data management, data integration, and analytics infrastructure supporting financial market oversight. In his current role, he is responsible for advancing the institution’s analytical capabilities through the implementation of modern data architectures, including the adoption of Data Mesh principles to overcome silos and foster cross-organizational data accessibility. His previous experience includes over decade in the largest bank in Central and Eastern Europe, where he led data quality and reporting teams. He holds an Executive MBA from the University of Warsaw and a Master of Science in Engineering from Bialystok University of Technology. At DEXA 2025, he presents a case study on the ongoing architectural transformation within a major supervisory institution, offering practical insights into designing scalable, federated data systems for public sector organizations.