arXiv subjects

Uraz Odyurt

Publications and source records attributed to Uraz Odyurt.

15 recordsLinked to original sources

arXiv2026

Recent Developments in Transformer Inference Deployment on FPGA Platforms: A Survey

With the rapid and continuous growth in the incorporation of machine learning models based on the Transformer architecture, capable deployment is in high demand. In this context, capable deployment refers to operational performance aspects, e.g., throughput and latency, as well as efficiency aspects, e.g., energy consumption. When it comes to the task of inference using such models, purpose-built hardware accelerators provide a lucrative alternative to common deployment choices, such as Central Processing Units (CPUs) and Graphics Processing Units (GPUs). The Field Programmable Gate Array (FPGA) platforms category is an example of such alternative accelerators, promising implementation flexibility, energy efficiency, improved latency and suitability for on-site deployment. We investigate the most recent advances, trends, and design choices for Transformer inference on FPGA platforms. We perform a systematic literature review, extracting and delving into preferred techniques for implementation and optimisation. This study and the provided taxonomy of topics could act as a guide for researchers from the academia and industry alike.

cs.LG↗

arXiv2026

MergeOver: Post-Training Token Merging for Recursive Vision Transformers

Vision Transformers (ViTs) demonstrate exceptional performance in computer vision but suffer from large parameter counts and quadratic computational complexity, severely limiting their deployment on resource-constrained edge hardware. While recursive weight-sharing reduces parameter counts and token merging mitigates computational and memory bottlenecks, integrating these two paradigms without costly retraining is non-trivial, leaving this intersection largely unexplored. We propose MergeOver, a post-training approach that integrates Token Merging (ToMe) into the recursively weight-shared Sliced Recursive Transformer (SReT). Through an Unmerge tracking stack, constraint-safe merge-rate adjustment, and synchronised token-mass tracking across spatial permutations, MergeOver resolves the spatial and merging constraints of this integration. We further employ a stage-wise single-shot schedule that performs token reduction at the first block of each stage and maintains a fixed sequence length throughout its subsequent recursive iterations. Benchmarked on ImageNet-1K, our selected configuration reduces top-1 accuracy by 1.47 percentage points. On the GPU, it reduces peak activation memory by 37.3% and 38.4% at batch sizes 1 and 16, while throughput decreases by 21.7% at batch size 1 but increases by 21.7% at batch size 16. On a Raspberry Pi 5 (ARM CPU), it reduces latency by 2.4% and 17.6% at batch sizes 1 and 16. These results show that MergeOver can recover a meaningful part of the throughput and memory cost that recursive weight-sharing introduces, without retraining, and provides a baseline for combining token merging with hierarchical recursive transformers.

cs.CV↗

arXiv2026

Implementing CPSLint: A Data Validation and Sanitisation Tool for Industrial Cyber-Physical Systems

Raw datasets are often too large and unstructured to work with directly, and require a data preparation phase. The domain of industrial Cyber-Physical Systems (CPSs) is no exception, as raw data typically consists of large time-series data collections that log the system's status at regular time intervals. The processing of such raw data is often carried out using ad hoc, case-specific, one-off Python scripts, often neglecting aspects of readability, reusability, and maintainability. In practice, this can cause professionals such as data scientists to write similar data preparation scripts for each case, requiring them to do much repetitive work. We introduce CPSLint, a Domain-Specific Language (DSL) designed to support the data preparation process for industrial CPS. CPSLint raises the level of abstraction to the point where both data scientists and domain experts can perform the data preparation task. We leverage the fact that many raw data collections in the industrial CPS domain require similar actions to render them suitable for data-centric workflows. In our DSL one can express the data preparation process in just a few lines of code. CPSLint is a publicly available tool applicable for any case involving time-series data collections in need of sanitisation.

cs.PL↗

arXiv2026

Explainable AI to Improve Machine Learning Reliability for Industrial Cyber-Physical Systems

Industrial Cyber-Physical Systems (CPS) are sensitive infrastructure from both safety and economics perspectives, making their reliability critically important. Machine Learning (ML), specifically deep learning, is increasingly integrated in industrial CPS, but the inherent complexity of ML models results in non-transparent operation. Rigorous evaluation is needed to prevent models from exhibiting unexpected behaviour on future, unseen data. Explainable AI (XAI) can be used to uncover model reasoning, allowing a more extensive analysis of behaviour. We apply XAI to improve predictive performance of ML models intended for an industrial CPS use-case. We analyse the effects of components from time-series data decomposition on model predictions using SHAP values. Through this method, we observe evidence on the lack of sufficient contextual information during model training. By increasing the window size of data instances, informed by the XAI findings for this use-case, we are able to improve model performance.

cs.LG↗

arXiv2025

Defining Energy Indicators for Impact Identification on Aerospace Composites: A Structured Feature Selection Approach Guided by Domain Knowledge

Energy estimation is critical to impact identification on aerospace composites, where low-velocity impacts can induce internal damage that is undetectable at the surface. Data sparsity, signal noise, complex feature interdependencies, non-linear dynamics, massive design spaces, and the ill-posed nature of the inverse problem often constrain current methodologies for energy prediction. Machine learning enriched with prior knowledge is a promising direction for overcoming these constraints. Prior knowledge can be incorporated by acting on the input space, where the choice of data representation directly influences how effectively the model relates measured signals to impact energy. Despite its importance, the selection of effective features lacks a systematic procedure, with no consensus on how to choose among the many candidate descriptors available. The present study addresses that gap through a structured workflow that designs the input space using domain knowledge. Features are extracted from the time, frequency, and time-frequency domains, then filtered for statistical significance, correlation, dimensionality reduction, and robustness to noise. Exploratory data analysis further relates the retained descriptors to the dynamics. The resulting indicators form the input space for a fully connected neural network, which is trained and validated on experimental data from multiple impact scenarios, including pristine and damaged states. The model reduces the prediction error by a factor of three relative to conventional time-series techniques and purely data-driven baselines, while every retained or discarded descriptor remains traceable to the aspect it describes. Overall, the framework advances predictive performance, interpretability, and diagnostic confidence by embedding domain knowledge through targeted feature selection.

cs.LG↗

arXiv2025

CPSLint: A Domain-Specific Language Providing Data Validation and Sanitisation for Industrial Cyber-Physical Systems

Industrial cyber-physical systems generate vast amounts of semi-structured time-series data that require careful preprocessing before they can be effectively used for machine learning applications such as fault detection and identification. Raw sensor datasets are often corrupted or incomplete, making it challenging to develop reliable solutions without proper data preparation and validation. In this paper, we introduce CPSLint, a domain-specific language for data validation and sanitisation. We present the design, implementation and evaluation of CPSLint, demonstrating its ability to automatically detect and correct common data corruption patterns while enabling non-programming domain experts to effectively prepare their data for analysis. We report evaluation results on a representative dataset, tracking memory consumption and CPU-time for sanitisation activities. Our approach offers several advantages over traditional methods, including reduced manual effort, guaranteed consistency and broader applicability across time-series datasets and projects.

cs.PL↗

arXiv2025

Demonstrators for Industrial Cyber-Physical System Research: A Requirements Hierarchy Driven by Software-Intensive Design

One of the challenges apparent in the organisation of research projects is the uncertainties around the subject of demonstrators. A precise and detailed elicitation of the coverage for project demonstrators is often an afterthought and not sufficiently detailed during proposal writing. This practice leads to continuous confusion and a mismatch between targeted and achievable demonstration of results, hindering progress. The reliance on the TRL scale as a loose descriptor does not help either. We propose a demonstrator requirements elaboration framework aiming to evaluate the feasibility of targeted demonstrations, making realistic adjustments, and assist in describing requirements. In doing so, we define 5 hierarchical levels of demonstration, clearly connected to expectations, e.g., work package interaction, and also connected to the project's industrial use-cases. The considered application scope in this paper is the domain of software-intensive systems and industrial cyber-physical systems. A complete validation is not accessible, as it would require application of our framework at the start of a project and observing the results at the end, taking 4-5 years. Nonetheless, we have applied it to two research projects from our portfolio, one at the early and another at the final stages, revealing its effectiveness.

cs.SE↗

arXiv2025

TrackCore-F: Deploying Transformer-Based Subatomic Particle Tracking on FPGAs

The Transformer Machine Learning (ML) architecture has been gaining considerable momentum in recent years. In particular, computational High-Energy Physics tasks such as jet tagging and particle track reconstruction (tracking), have either achieved proper solutions, or reached considerable milestones using Transformers. On the other hand, the use of specialised hardware accelerators, especially FPGAs, is an effective method to achieve online, or pseudo-online latencies. The development and integration of Transformer-based ML to FPGAs is still ongoing and the support from current tools is very limited or non-existent. Additionally, FPGA resources present a significant constraint. Considering the model size alone, while smaller models can be deployed directly, larger models are to be partitioned in a meaningful and ideally, automated way. We aim to develop methodologies and tools for monolithic, or partitioned Transformer synthesis, specifically targeting inference. Our primary use-case involves two machine learning model designs for tracking, derived from the TrackFormers project. We elaborate our development approach, present preliminary results, and provide comparisons.

hep-ex↗

arXiv2025

TrackFormers Part 2: Enhanced Transformer-Based Models for High-Energy Physics Track Reconstruction

High-Energy Physics experiments are rapidly escalating in generated data volume, a trend that will intensify with the upcoming High-Luminosity LHC upgrade. This surge in data necessitates critical revisions across the data processing pipeline, with particle track reconstruction being a prime candidate for improvement. In our previous work, we introduced "TrackFormers", a collection of Transformer-based one-shot encoder-only models that effectively associate hits with expected tracks. In this study, we extend our earlier efforts by conducting detailed investigations into more custom Transformer attention mechanisms, a new design combining geometric projection and lightweight clustering, and a joint model conditioning classification on a regressor's predictions. Furthermore, we discuss new datasets that allow the training on hit level for a range of physics processes. These developments collectively aim to boost both the accuracy and potentially the efficiency of our tracking models, offering a robust solution to meet the demands of next-generation high-energy physics experiments.

hep-ex↗

arXiv2025

InfoPos: A Design Support Framework for ML-Assisted Fault Detection and Identification in Industrial Cyber-Physical Systems

The variety of building blocks and algorithms incorporated in data-centric and ML-assisted fault detection and identification solutions is high, contributing to two challenges: selection of the most effective set and order of building blocks, as well as achieving such a selection with minimum cost. Considering that ML-assisted solution design is influenced by the extent of available data and the extent of available knowledge of the target system, it is advantageous to be able to select effective and matching building blocks. We introduce the first iteration of our InfoPos framework, allowing the placement of fault detection/identification use-cases based on the available levels (positions), i.e., from poor to rich, of knowledge and data dimensions. With that input, designers and developers can reveal the most effective corresponding choice(s), streamlining the solution design process. The results from a demonstrator, a fault identification use-case for industrial Cyber-Physical Systems, reflects achieved effects when different building blocks are used throughout knowledge and data positions. The achieved ML model performance is considered as the indicator for a better solution. The data processing code and composed datasets are publicly available.

cs.LG↗

arXiv2024

TrackFormers: In Search of Transformer-Based Particle Tracking for the High-Luminosity LHC Era

High-Energy Physics experiments are facing a multi-fold data increase with every new iteration. This is certainly the case for the upcoming High-Luminosity LHC upgrade. Such increased data processing requirements forces revisions to almost every step of the data processing pipeline. One such step in need of an overhaul is the task of particle track reconstruction, a.k.a., tracking. A Machine Learning-assisted solution is expected to provide significant improvements, since the most time-consuming step in tracking is the assignment of hits to particles or track candidates. This is the topic of this paper. We take inspiration from large language models. As such, we consider two approaches: the prediction of the next word in a sentence (next hit point in a track), as well as the one-shot prediction of all hits within an event. In an extensive design effort, we have experimented with three models based on the Transformer architecture and one model based on the U-Net architecture, performing track association predictions for collision event hit points. In our evaluation, we consider a spectrum of simple to complex representations of the problem, eliminating designs with lower metrics early on. We report extensive results, covering both prediction accuracy (score) and computational performance. We have made use of the REDVID simulation framework, as well as reductions applied to the TrackML data set, to compose five data sets from simple to complex, for our experiments. The results highlight distinct advantages among different designs in terms of prediction accuracy and computational performance, demonstrating the efficiency of our methodology. Most importantly, the results show the viability of a one-shot encoder-classifier based Transformer solution as a practical approach for the task of tracking.

hep-ex↗

arXiv2024

Novel Approaches for ML-Assisted Particle Track Reconstruction and Hit Clustering

Track reconstruction is a vital aspect of High-Energy Physics (HEP) and plays a critical role in major experiments. In this study, we delve into unexplored avenues for particle track reconstruction and hit clustering. Firstly, we enhance the algorithmic design effort by utilising a simplified simulator (REDVID) to generate training data that is specifically composed for simplicity. We demonstrate the effectiveness of this data in guiding the development of optimal network architectures. Additionally, we investigate the application of image segmentation networks for this task, exploring their potential for accurate track reconstruction. Moreover, we approach the task from a different perspective by treating it as a hit sequence to track sequence translation problem. Specifically, we explore the utilisation of Transformer architectures for tracking purposes. Our preliminary findings are covered in detail. By considering this novel approach, we aim to uncover new insights and potential advancements in track reconstruction. This research sheds light on previously unexplored methods and provides valuable insights for the field of particle track reconstruction and hit clustering in HEP.

hep-ex↗

arXiv2024

Model Parallelism on Distributed Infrastructure: A Literature Review from Theory to LLM Case-Studies

Neural networks have become a cornerstone of machine learning. As the trend for these to get more and more complex continues, so does the underlying hardware and software infrastructure for training and deployment. In this survey we answer three research questions: "What types of model parallelism exist?", "What are the challenges of model parallelism?", and "What is a modern use-case of model parallelism?" We answer the first question by looking at how neural networks can be parallelised and expressing these as operator graphs while exploring the available dimensions. The dimensions along which neural networks can be parallelised are intra-operator and inter-operator. We answer the second question by collecting and listing both implementation challenges for the types of parallelism, as well as the problem of optimally partitioning the operator graph. We answer the last question by collecting and listing how parallelism is applied in modern multi-billion parameter transformer networks, to the extend that this is possible with the limited information shared about these networks.

cs.DC↗

arXiv2024

Reduced Simulations for High-Energy Physics, a Middle Ground for Data-Driven Physics Research

Subatomic particle track reconstruction (tracking) is a vital task in High-Energy Physics experiments. Tracking is exceptionally computationally challenging and fielded solutions, relying on traditional algorithms, do not scale linearly. Machine Learning (ML) assisted solutions are a promising answer. We argue that a complexity-reduced problem description and the data representing it, will facilitate the solution exploration workflow. We provide the REDuced VIrtual Detector (REDVID) as a complexity-reduced detector model and particle collision event simulator combo. REDVID is intended as a simulation-in-the-loop, to both generate synthetic data efficiently and to simplify the challenge of ML model design. The fully parametric nature of our tool, with regards to system-level configuration, while in contrast to physics-accurate simulations, allows for the generation of simplified data for research and education, at different levels. Resulting from the reduced complexity, we showcase the computational efficiency of REDVID by providing the computational cost figures for a multitude of simulation benchmarks. As a simulation and a generative tool for ML-assisted solution design, REDVID is highly flexible, reusable and open-source. Reference data sets generated with REDVID are publicly available. Data generated using REDVID has enabled rapid development of multiple novel ML model designs, which is currently ongoing.

hep-ex↗

arXiv2024

Towards Seamless Serverless Computing Across an Edge-Cloud Continuum

Serverless computing has emerged as an attractive paradigm due to the efficiency of development and the ease of deployment without managing any underlying infrastructure. Nevertheless, serverless computing approaches face numerous challenges to unlock their full potential in hybrid environments. To gain a deeper understanding and firsthand knowledge of serverless computing in edge-cloud deployments, we review the current state of open-source serverless platforms and compare them based on predefined requirements. We then design and implement a serverless computing platform with a novel edge orchestration technique that seamlessly deploys serverless functions across the edge and cloud environments on top of the Knative serverless platform. Moreover, we propose an offloading strategy for edge environments and four different functions for experimentation and showcase the performance benefits of our solution. Our results demonstrate that such an approach can efficiently utilize both cloud and edge resources by dynamically offloading functions from the edge to the cloud during high activity, while reducing the overall application latency and increasing request throughput compared to an edge-only deployment.

cs.DC↗