Fast Machine Learning for Science Conference 2025

Session

Contributed talks

2 Sept 2025, 13:00

ETH Zurich

103. FINN+: Towards Hassle-Free Co-Design of FPGA DNN Inference Accelerators

Felix Jentzsch

02/09/2025, 13:00

Standard Talk

Custom FPGA dataflow accelerators for DNN inference can enable unprecedented performance and efficiency for many applications. Dataflow accelerator compilers, such as the FINN framework, have improved in recent years and allow practitioners to explore this technology without requiring in-depth FPGA knowledge.

However, the overall design process remains quite tedious, time-consuming, and...

22. End-to-End Neural Network Compression and Deployment for Hardware Acceleration Using PQuant and hls4ml

Roope Oskari Niemi

02/09/2025, 13:20

Standard Talk

As the demand for efficient machine learning on resource-limited devices grows, model compression techniques like pruning and quantization have become increasingly vital. Despite their importance, these methods are typically developed in isolation, and while some libraries attempt to offer unified interfaces for compression, they often lack support for deployment tools such as hls4ml. To...

99. PrioriFI: Efficient Fault Injection for Edge Neural Networks

Olivia Weng

02/09/2025, 13:40

Standard Talk

As neural networks (NNs) are increasingly used to provide
edge intelligence, there is a growing need to make the edge devices
that run them robust to faults. Edge devices must mitigate the resulting
hardware failures while maintaining strict constraints on power, energy,
latency, throughput, memory size, and computational resources. Edge
NNs require fundamental changes in model...

107. ENABOL: Enabling Neural Backpropagation On-chip Learning for Edge AI Systems

Manuel Valentin (Northwestern University)

02/09/2025, 14:00

Standard Talk

On-chip learning has the potential to unlock low-latency, low-power, and continuously adaptive AI directly on edge devices. However, research in this area remains limited by the lack of accessible hardware toolchains that support backpropagation. To address this gap, we propose ENABOL, a hardware-efficient extension of the HLS4ML toolchain that enables customizable backpropagation support...

129. da4ml: Distributed Arithmetic for Real-time Neural Networks on FPGAs

Chang Sun (California Institute of Technology (US))

02/09/2025, 14:20

Standard Talk

Neural networks with a latency requirement on the order of microseconds, like the ones used at the CERN Large Hadron Collider, are typically deployed on FPGAs pipelined with II=1. A bottleneck for the deployment of such neural networks is area utilization, which is directly related to the required constant matrix-vector multiplication (CMVM) operations. In this work, we propose an efficient...

31. AMD AI-Engines in fixed latency environments

Ioannis Xiotidis (CERN)

02/09/2025, 14:40

Standard Talk

The ATLAS Level-0 Global Trigger is a mission critical system opting to take advantage of the full calorimeter granularity during Run-4 and beyond. Level-0 Global will be executing a cascade of trigger algorithms combined both the calorimeter information and the muons. Within the Next Generation Trigger project at CERN there is a dedicated work package (WP2.1) exploring large deployment of...

139. Balancing Prediction Performance, Transparency and Energy Consumption in Machine Learning Models for Data Streams

Kirsten Köbschall

02/09/2025, 16:00

Standard Talk

In the era of continuous data generation, real-time processing of data streams has become crucial for timely, adaptive, and context-aware decision-making. However, maintaining effective learning models in such dynamic environments requires carefully balancing prediction performance, transparency and energy consumption.

In the talk, we will present two new state-of-the-art methods for...

113. Arbolta: A Fault Tolerance Study using Minimal Hardware Simulation

Alexander Redding (UC San Diego)

02/09/2025, 16:20

Standard Talk

The widespread deployment of embedded ML systems has created a need for resilient, fault-tolerant hardware and software capable of operating in inherently noisy conditions. While the standardization of low-precision (≤ 8-bit) datatypes has allowed for reduced training and inference costs and increased interoperability across commercial accelerators, clear guidelines for robust implementation...

108. SuperSONIC: Cloud-Native Infrastructure for ML Inferencing

Yuan-Tang Chou (University of Washington (US))

02/09/2025, 16:40

Standard Talk

The rising computational demands of increasing data rates and complex machine learning (ML) algorithms in large-scale scientific experiments have driven the adoption of the Services for Optimized Network Inference on Coprocessors (SONIC) framework. SONIC accelerates ML inference by offloading tasks to local or remote coprocessors, optimizing resource utilization. Its portability across diverse...

90. Embedding domain knowledge: Inductive biases for algorithmic alignment in Machine Learning

Serio Angelo Maria Agriesti (Department of Technology, Management and Economics, Technical University of Denmark, Lyngby, Denmark)

02/09/2025, 17:00

Standard Talk

Most of the current machine learning (ML) applications are purely data-driven solutions with little considerations for underlying problem dynamics, limited to in-distribution applications. To tackle this limitation a stream of literature is emerging to address out-of-distribution (OOD) performance: Algorithmic alignment, which focuses on embedding algorithmic structures into ML architectures...

38. Pushing Matrix-Vector Multiplication Performance on AMD AI Engines for Low-Latency Edge Inference

Dimitrios Danopoulos (CERN)

02/09/2025, 17:20

Standard Talk

Matrix-vector (GEMV) operations are a common building block in many deep learning models, particularly for large dense layers found in convolutional neural networks (CNNs) and multi-layer perceptrons (MLPs). Despite their importance, GEMV kernels have historically underperformed compared to matrix-matrix (GEMM) operations due to their lower arithmetic intensity and limited data reuse, making...

65. Real-Time Anomaly Detection in the CMS Level-1 Trigger with AXOL1TL

Sabrina Giorgetti (Universita e INFN, Padova (IT))

03/09/2025, 13:00

Standard Talk

AXOL1TL is an anomaly detection (AD) trigger algorithm integrated into the Global Trigger (GT) of the CMS Level-1 Trigger (L1T) system since 2024. The GT reduces the event rate from proton–proton collisions at the LHC, lowering it from 40 MHz to 100 kHz within a fixed latency of 50 ns. The AD algorithm, implemented in the FPGA firmware of the GT board, uses an autoencoder to assign an anomaly...

85. GELATO: A Generic Event-Level Anomaly detection Trigger for ATLAS

Kenny Jia (Stanford University/ SLAC)

03/09/2025, 13:20

Contributed Talks

Standard Talk

The absence of BSM physics discoveries at the LHC suggests new physics could lie outside current trigger schemes. By applying unsupervised ML–based anomaly detection, we gain a model-agnostic way of spotting anomalous signatures that deviate from the current trigger’s expectations. Here we introduce a Run-3 trigger chain that embeds fast anomaly detection algorithms in both hardware and...

35. Advancing the CMS Level-1 Trigger: Jet Tagging with DeepSets at the HL-LHC

Christopher Edward Brown (CERN)

03/09/2025, 13:40

Standard Talk

At the Phase-2 Upgrade of the CMS Level-1 Trigger (L1T), particles will be reconstructed by linking charged particle tracks with clusters in the calorimeters and muon tracks from the muon station. The 200 pileup interactions will be mitigated using primary vertex reconstruction for charged particles and a weighting for neutral particles based on the distribution of energy in a small area. Jets...

84. Low-Latency On-Chip Tau Event Selection with Machine Learning for the Belle II Level-1 Trigger

Deven Misra (University of Tokyo)

03/09/2025, 14:00

Standard Talk

Belle II is a luminosity frontier experiment located at the SuperKEKB asymmetric $e^+ e^-$ collider, operating at the $\Upsilon(4S)$ resonance. The $\tau$ physics program at Belle II involves both probes of new physics and precision measurements of standard model parameters with large statistics. SuperKEKB is projected to reach a luminosity of $6\times 10^{35}~\text{cm}^{-2}\text{s}^{-1}$ in...

93. Accelerated Graph Neural Network Inference on FPGAs for Real-Time Muon Triggering at the HL-LHC

Davide Fiacco (Sapienza Universita e INFN, Roma I (IT))

03/09/2025, 14:20

Standard Talk

The High Luminosity upgrade of the Large Hadron Collider (HL-LHC) presents a demanding environment for real-time data processing, with substantially increased event rates requiring faster and more efficient trigger systems. This study explores the deployment of graph neural networks (GNNs) on field-programmable gate arrays (FPGAs) for fast and accurate inference within future muon trigger...

27. Jet finding in real-time using an object detection CNN

Leon Bozianu (Universite de Geneve (CH))

03/09/2025, 14:40

Standard Talk

The ATLAS trigger system will undergo a comprehensive upgrade in advance of the HL-LHC programme. In order to deal with the increased data bandwidth trigger algorithms will be required to satisfy stricter latency requirements. We propose a method to speed up the current calorimeter-only preselection step and to aid trigger decisions for hadronic signals containing jets.
We demonstrate the use...

134. KAN-LUT: Efficient LUT-Based Acceleration of Kolomogorov-Arnold Networks (KANs) on FPGAs

Duc Hoang (Massachusetts Inst. of Technology (US))

03/09/2025, 16:00

Standard Talk

Optimized FPGA implementations of tiny neural networks are crucial for low-latency and hardware-efficient inference for a variety of applications. Neural networks based on lookup tables (LUTs) are a standard technique for such problems due to their hardware efficiency and strong expressivity. However, such networks are often difficult to scale up as their resource usage scales exponentially...

133. COLLIDE-2V - 750 Million Dual-View LHC Event Dataset for Low-Latency ML

Eric Anton Moreno (Massachusetts Institute of Technology (US))

03/09/2025, 16:20

Standard Talk

Modern foundation models (FMs) have pushed the frontiers of language, vision, and multi-model tasks by training ever-larger neural networks (NN) on unprecedented volumes of data. The use of FM models has yet to be established in collider physics, which both lack a comparably sized, general-purpose dataset on which to pre-train universal event representations, and a clear demonstrable need....

128. Accelerating Efficient Transformer Architectures for Point Cloud Data using hls4ml (REMOTE)

Jan-Frederik Schulte (Purdue University (US))

03/09/2025, 16:40

Standard Talk

The analysis of point cloud data, for example signals from charged particles recorded by detectors in high energy physics (HEP) experiments, can be significantly enhanced and accelerated by the application of machine learning models. In recent years, transformer architectures have come into focus as offering excellent model performance. However, for traditional transformers,the need to compute...

86. Hierarchical Dataflow Accelerator of Interaction Networks for Large-Scale Particle Tracking on FPGA

Bo-Cheng Lai

03/09/2025, 17:00

Standard Talk

The Interaction Network (IN) algorithm has shown great promise for particle tracking applications at the Large Hadron Collider (LHC), where identifying complex particle trajectories from raw detector data is a computationally intensive task. IN leverages graph-based representations of detector hits to learn relationships between particle interactions, making it well-suited for this domain....

159. An ML Pipeline for Real-time Gravitational Wave Alerts

Christina Reissel (Massachusetts Inst. of Technology (US)), Katya Govorkova (Massachusetts Inst. of Technology (US)), Philip Coleman Harris (Massachusetts Inst. of Technology (US))

03/09/2025, 17:20

43. Towards a Self-Driving Trigger: Adaptive Response in Real Time

Giovanna Salvi (University of Michigan (US))

04/09/2025, 11:30

Standard Talk

The trigger systems of ATLAS and CMS currently reject vast numbers of potentially valuable collision events due to their conservative, static designs, a limitation that directly hampers discovery potential. We propose an alternative to these rigid, hand-tuned menus with an autonomous controller capable of dynamically optimizing trigger performance in real time.
In this work, we demonstrate...

37. Easing the path to deployment in ML4Sys through FPGAs

Maximilian Heer (ETH Zurich)

04/09/2025, 11:50

Standard Talk

Machine Learning (ML) techniques are increasingly applied for the optimization of complex computing systems, but their integration into core low-level system mechanisms remains limited. A key barrier is the lack of accessible, high- performance interfaces at the boundary between software and hardware as well as hardware-offloaded ML-inference at full systems speed. In this presentation, we...

32. Go small then go home - hyperparameter transfer for ML in HEP

Liv Helen Vage (Princeton University (US))

04/09/2025, 12:10

Standard Talk

Tuning hyperparameters of ML models, especially large ML models, can be time consuming and computationally expensive. As a potential solution, several recent papers have explored hyperparameter transfer. Under certain conditions, the optimal hyperparameters of a small model are also optimal for larger models. One can therefore tune only the small model and transfer the hyperparameters to the...

160. Edge Deep Learning for Particle Physics (EPIGRAPHY)

Benedikt Maier (Imperial College (GB))

04/09/2025, 12:30

102. Low-Latency Resource-Efficient GNNs for Jet Tagging on FPGAs

Zhiqiang (Walkie) Que (Imperial College London)

04/09/2025, 13:30

Standard Talk

Graph Neural Networks (GNNs), particularly Interaction Networks (INs), have shown exceptional performance for jet tagging at the CERN High-Luminosity Large Hadron Collider. However, their computational complexity and irregular memory access patterns pose significant challenges for deployment on FPGAs in hardware trigger systems, where strict latency and resource constraints apply.

In this...

117. Smartpixels: Intelligent pixel detectors: Towards a radiation hard ASIC with on-chip machine learning in 28nm CMOS

Benjamin Weiss (Cornell University), Jannicke Pearkes (University of Colorado Boulder (US))

04/09/2025, 13:50

Standard Talk

The Smartpixels project is a coordinated effort to co-design pixel ASICs, design tools, ML algorithms, and sensors for on-detector data reduction, motivated by the technical challenges of current and future colliders. The drive to greater precision requires smaller pixel pitch, which together with higher event rates arising from pileup and/or beam-induced background generates petabytes of data...

96. Quantum-Inspired Tensor Network Models for Ultrafast Jet Tagging on FPGAs

Ms Ema Puljak (universitat Autònoma de Barcelona)

04/09/2025, 14:10

Standard Talk

We conduct a systematic study of quantum-inspired Tensor Network (TN) models—Matrix Product States (MPS) and Tree Tensor Networks (TTN)—for real-time jet tagging in high-energy physics, with a focus on low-latency deployment on FPGAs. Motivated by the strict computational demands of the HL-LHC Level-1 Trigger system, we explore TN architectures as compact and interpretable alternatives to deep...

88. Toward a Ultra-Fast, Energy-Efficient Readout of Calorimeters with Neuromorphic Processing

Enrico Lupi (CERN, INFN Padova (IT))

04/09/2025, 14:30

Standard Talk

Hadronic calorimeters are a key part of high energy physics experiments. Traditionally, they rely on high granularity to improve performances, but this leads to various challenges in terms of cost, energy consumption and output data volume. Moreover, current detectors do not have the capability of exploiting temporal information of the shower development, as the time frame for pattern...

79. SparsePixels: Efficient Convolution for Sparse Data on FPGAs

Ho-Fung Tsoi (University of Pennsylvania)

04/09/2025, 14:50

Standard Talk

Inference of standard convolutional neural networks (CNNs) on FPGAs often incurs high latency and long initiation intervals due to the nested loops required to slide filters across the full input, especially when the input dimensions are large. However, in some datasets, meaningful signals may occupy only a small fraction of the input, say sometimes just a few percent of the total pixels or...

73. FPGA-accelerated ML for real-time RHEED inference

Abdelrahman Asem Elabd (University of Washington (US))

04/09/2025, 15:10

Standard Talk

Reflection High-Energy Electron Diffraction (RHEED) is a common diffraction-based surface characterization technique for analyzing the properties of crystalline materials that are grown using a thin-film deposition technique like pulsed-laser deposition (PLD) or molecular-beam epitaxy (MBE). In this work, we design an FPGA-accelerated machine learning (ML) algorithm to perform real-time...

121. Low-latency Jet Tagging for HL-LHC Using Transformer Architectures

Lauri Antti Olavi Laatu (Imperial College (GB))

04/09/2025, 16:00

Standard Talk

Transformers are the state-of-the-art model architectures and widely used in application areas of machine learning. However the performance of such architectures is less well explored in the ultra-low latency domains where deployment on FPGAs or ASICs is required. Such domains include the trigger and data acquisition systems of the LHC experiments.

We present a transformer-based algorithm...

21. Radiation-Hard, ML-Based, Low-Latency Compression for the LHCb ECAL Upgrade

Katya Govorkova (Massachusetts Inst. of Technology (US))

04/09/2025, 16:20

Standard Talk

The LHCb Upgrade II will operate at a data rate of 200 Tb/s, requiring efficient real-time data reduction. A major challenge of this pipeline is the transfer of full timing information from the frontend Electromagnetic Calorimeter (ECAL) to the backend for processing, which is critical for resolving pile-up, background suppression, and enhancing energy resolution. Due to the data rate, full...

51. Chisel4ml: Using Chisel For Direct Circuit Implementation of Deeply Quantized Neural Networks

Jure Vreča

04/09/2025, 16:40

Poster

We give an introduction to chisel4ml, a tool for generating direct circuit implementations of deeply quantized neural networks. It uses structural descriptions of deeply quantized neural networks in the form of Chisel generators. Chisel is a domain-specific language for designing synchronous digital circuits. It is a language embedded in Scala that offers a wealth of powerful features, such...

54. A Real-Time GNN-based Clustering Algorithm for the Level 1 Calorimeter Trigger at Belle II

Isabel Haide (Karlsruhe Institute for Technology)

05/09/2025, 10:15

Standard Talk

With increasing beam background levels at Belle II, which have already been observed due to the world-record instantaneous luminosities achieved by SuperKEKB and which are expected to rise further, an upgrade of the current Level 1 (L1) trigger algorithms is necessary to handle the evolving conditions. In this work, we present an upgraded L1 electromagnetic calorimeter trigger, based on Graph...

58. Experiences Deploying a Hybrid PVFinder Algorithm for Primary Vertex Reconstruction in LHCb’s GPU-Resident HLT1

Mohamed Elashri (University of Cincinnati)

05/09/2025, 10:35

Standard Talk

The PVFinder algorithm employs a hybrid deep neural network (DNN) approach to reconstruct primary vertices (PVs) in proton-proton collisions at the LHC, addressing the complexities of high pile-up environments in LHCb and ATLAS experiments. By integrating fully connected layers with a UNet architecture, PVFinder’s end-to-end tracks-to-hist DNN processes charged track parameters to predict PV...

150. State space models for Project 8 event reconstruction

Hannah Binney

05/09/2025, 10:55

Contributed Talks

Standard Talk

The Project 8 experiment aims to directly probe the neutrino mass by precisely measuring the energy spectrum of beta electrons emitted in the decay of tritium. The collaboration has pioneered the cyclotron radiation emission spectroscopy technique (CRES), which measures the energy of single electrons by detecting the cyclotron radiation they emit in a magnetic field. Traditional methods for...

163. Get your poster!

05/09/2025, 11:15

92. FPGA-Based Quantized Inception-Style Neural Networks for Enhancing Spatial Resolution in Electron Microscopy

Lorenzo Asfour (ETH Zurich (CH))

Standard Talk

High-resolution electron microscopy generates large volumes of pixel detector data due to beam rates reaching $10^7$ to $10^{10}$ electrons per second directed at the sample. Of this data, only the electron entry point into the silicon detector prior to scattering is typically of interest for downstream analysis. Precise knowledge of these entry points is particularly important in electron...