Student Projects

Vision Transformers (ViTs) demonstrate exceptional performance in computer vision but suffer from large parameter counts and quadratic computational complexity, $\mathcal{O}\left(N^2\right)$, 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 a post-training multi-axis compression approach that successfully combines the recursive weight-sharing of the Sliced Recursive Transformer (SReT) with the dynamic token merging algorithm of Token Merging (ToMe). By implementing an Unmerge tracking stack, enforcing strict mathematical merging bounds, and applying parallel spatial tracking, our methodology resolves the spatial and merging constraints of the integration. Furthermore, we utilise an exponential token reduction schedule to stabilise the semantic densification inherent to recursive loops. Benchmarked on ImageNet-1K, our optimised configuration achieves a 27.6% increase in throughput and a 38.5% reduction in Peak Activation Memory (PAM) with a minimal 1.47% accuracy drop on a GPU at a batch size of 128. However, the algorithmic overhead negated the performance gains at a batch size of 1. Nevertheless, this approach establishes the feasibility of dynamic token reduction within recursive ViT architectures, providing a structural baseline for future edge-targeted optimisations.

The High-Luminosity Large Hadron Collider (HL-LHC) will significantly increase data processing requirements for tasks such as particle tracking. Transformer-based models have recently shown promising results for track reconstruction, but running such models in hardware environments relevant to high-energy physics is far from straightforward. Strict latency constraints, limited on-chip memory, and integration into existing systems all have to be considered.

This thesis explores how a Transformer-based tracking model can be deployed on an FPGA platform under realistic resource and accuracy constraints. Instead of aiming for a full end-to-end implementation, the focus is placed on accelerating a representative Transformer encoder layer and studying the practical trade-offs that arise in hardware. An FPGA accelerator prototype is developed using high-level synthesis, with particular attention to on-chip weight storage, memory organization, and various hardware optimizations.

Beyond the hardware design itself, this work also addresses the practical challenge that deploying machine learning models on FPGAs typically requires expertise in both deep learning and hardware design. To lower this barrier, a semi-automated workflow is developed that enables selective offloading Transformer model parts to the FPGA, while the remaining parts of the model execute on a general-purpose processor. The goal is not only to build an accelerator, but also to make experimentation with FPGA-based acceleration more accessible to researchers without extensive hardware experience.

Experimental results show that the design is largely memory-bound, with on-chip memory usage forming the main bottleneck. Multi-layer implementations improve throughput and energy efficiency, but the achieved latency remains higher than that of modern GPU solutions. Overall, the results demonstrate that selective acceleration of Transformer components on FPGA hardware is feasible and provide a practical, reproducible workflow that helps bridge the gap between machine learning models and FPGA deployment in a high-energy physics context.

Neural networks have become increasingly complex, driving the need for efficient execution. This need is particularly pronounced in environments where computational resources are limited and the size of the models poses a challenge to traditional execution methods. One way to address this is through model parallelism where we partition a model over multiple devices. This thesis presents a method for enhancing model parallelism in existing neural neural network architectures, specifically focusing on the Inception-ResNet-v2 model. Our method consist of 3 transformations: placing sequential cells in parallel (ParallelResNet), replacing regular convolutions with grouped convolutions (GroupedResNet), and splitting operators along the channel dimension (SplitResNet). We analyze our proposed transformations on the Inception-ResNet-v2 model and show that they introduce a straightforward way to partition the model that minimizes communication overhead and creates partitions of equal computation complexity. We conclude that the classification accuracy is reduced for ParallelResNet, and increases for GroupedResNet. SplitResNet is either unaffected or reduced depending on the dataset. However, overall the effect of the transformations on classification accuracy is very small and further evaluation on more complex classification problems is necessary to quantify it.

In recent years, serverless computing has emerged as a transformative paradigm for building and deploying applications without the burden of managing traditional server infrastructure. Concurrently, edge computing has gained traction as a critical solution to reduce latency and enhance real-time data processing by bringing computation closer to data sources. However, while integrating these two powerful concepts – serverless and edge computing – has historically posed challenges, recent advancements have showcased the feasibility of running serverless platforms within edge environments. This paper introduces Knative Edge, a novel serverless platform that seamlessly orchestrates the deployment of serverless functions across both edge devices and cloud environments. Knative Edge leverages the unique characteristics of edge computing, such as proximity to data and reduced network latency, while harnessing the scalability and flexibility of serverless computing. By combining these two paradigms, Knative Edge empowers developers to build highly responsive and efficient applications that span the cloud-edge continuum. We present the architectural design, key components, and deployment workflow of Knative Edge, and provide experimental results showcasing its performance benefits. Our results demonstrate that Knative Edge can make efficient use of both cloud and edge resources by dynamically offloading function executing from the edge to the cloud periods of high activity, while reducing the overall application latency and increasing request throughput compared to an edge-only deployment.

In recent years, Serverless computing has emerged as a persuasive paradigm aiming to reshape the cloud computing landscape considerably. Serverless offers a scalable and cost-effective deployment model where users can run applications without the need to manage or provision servers. The underlying infrastructure is entirely abstracted and has the ability to scale automatically in a flexible manner, while the users are charged exclusively for the resources they use. In parallel, we have witnessed a surge in the adoption of Artificial Intelligence and Machine Learning (ML) technologies in various application domains. Since Serverless architectures are not tailored to address the unique challenges posed by resource-intensive jobs, combining ML with Serverless proves to be a complex undertaking.

In this thesis, we propose a solution for deploying ML models on Serverless platforms, specifically for inference jobs. Our model-agnostic approach is based on a flexible decomposition of such models into sub- models, referred to as slices, and the execution of inferences in a workflow of Serverless functions. We rely on conducting a thorough investigation of the limitations affecting the most popular Serverless platforms on the market and devising strategies to overcome them. Our experimental evaluations are performed on AWS, considering the ONNX open source format for ML model representation. Our results show that our decomposition method enables running ML inference on Serverless, regardless of the model size, benefiting from the high scalability of this architecture while lowering the strain on computing resources such as required runtime memory.

No abstract yet

No abstract yet

No abstract yet

Simulation and synthetic data play a pivotal role in High-Energy Physics (HEP) research, offering physics-accurate but slow frameworks and faster alternatives balancing speed and accuracy. This project extends the REDVID simulation framework, incorporating features like interaction with Monte Carlo event generators, basic magnetic field effects, pile-up effects, and considerations for reproducibility. Aimed at facilitating Machine Learning (ML) model design research for particle track reconstruction at the HL-LHC, the enhanced REDVID enables the generation of training data.

Simulation and synthetic data are integral to High-Energy Physics (HEP) research, offering physics-accurate but computationally demanding frameworks. Parametric and complexity-aware simulation tools simplify complexities, generating manageable data sets. This project extends the REDVID simulation framework to incorporate a time dimension, crucial for tracking particle interactions in detectors. As part of ongoing efforts for Machine Learning (ML) model design for particle track reconstruction at the HL-LHC, the enhanced REDVID enables the generation of training data with time information, facilitating 4-dimensional tracking.

Simulation and synthetic data play a crucial role in High-Energy Physics (HEP) research, with physics-accurate frameworks offering realistic data but being computationally intensive. This project extends the REDVID simulation framework to generate data at various complexity levels, integrating features like spherical coordinates, sub-detector layers, and track randomisation protocols. By focusing on complexity-reduced simulations, the aim is to facilitate Machine Learning (ML) model design research for particle track reconstruction at the HL-LHC. The student will define simulation recipes for different complexity levels, contributing to the ongoing efforts in ML model development.

Simulation and synthetic data generation play a crucial role in High-Energy Physics (HEP) research, with physics-accurate frameworks providing realistic but slow data syntheses. Fast simulation frameworks balance speed and accuracy, enabling various applications, including their role in Machine Learning (ML) model design research. This project extends the REDVID simulation framework to support different particles, especially electrons, essential for accurately simulating particle behaviour. The addition includes implementing electron interactions with matter, bremsstrahlung radiation, and energy loss effects, with a focus on their impact on ML model training. The improved REDVID aids in generating training data for ML models aimed at particle track reconstruction for the HL-LHC, or investigations into the new features' impact on ML models.

Simulation and synthetic data are integral to High-Energy Physics (HEP) research, offering both physics-accurate but slow frameworks and faster, simplified alternatives. This project aims to enhance the REDVID simulation framework by incorporating key features, notably support for muons. Muons possess distinct characteristics such as higher penetration power, instability leading to decay, interaction with matter, and susceptibility to magnetic fields. The goal is to implement these traits into REDVID, facilitating the generation of training data for Machine Learning (ML) models aimed at particle track reconstruction for the HL-LHC. The student shall study, select and implement a minimum set of distinguishing characteristics to REDVID.

No abstract yet

No abstract yet

No abstract yet

No abstract yet

No abstract yet