Roel Wuyts

Doing high throughput high accuracy metrology in small geometries is challenging. One approach is to build easily measurable proxy targets onto dies and make a predictive model based on those signals. We use optical Pattern Shift Response (PSR) proxy targets to build predictive models of the electrical characteristics of devices in the Back End Of Line (BEOL). Given the wide choice of PSR targets, we explore how to select combinations of them to maximise the utility of the features for building an accurate Machine Learning (ML) model; we call this approach Multiplexed Optical Metrology. We also explore the trade-off between chip area dedicated to targets and achievable accuracy. We run ML experiments using different selections of targets measured at different stages of BEOL processing: post-lithography and post-Chemical-Mechanical-Planarisation (CMP). Our results show that a) reasonable predictive performance can be achieved for a reasonable area budget; b) ML model performance across target families varies significantly, thus justifying the need for careful selection of targets; c) longitudinal measurements of targets increases accuracy for no extra area penalty; d) increasing the the number of targets gives some improvement in accuracy for a dataset of this size, but relatively small compared to the increase in area budget needed. Ultimately we aim to do die-level yield prediction using these techniques. We discuss how collecting a larger dataset with appropriate yield information is the logical next step to achieving this.

Background: Despite apparent promise and the availability of numerous examples in the literature, machine learning models are rarely used in practice in ICU units. This mismatch suggests that there are poorly understood barriers preventing uptake, which we aim to identify. Methods: We begin with a qualitative study with 29 interviews of 40 Intensive Care Unit-, hospital- and MedTech company staff members. As a follow-up to the study, we attempt to quantify some of the technical issues raised. To perform experiments we selected two models based on criteria such as medical relevance. Using these models we measure the loss of performance in predictive models due to drift over time, change of available patient features, scarceness of data, and deploying a model in a different context to the one it was built in. Results: The qualitative study confirms our assumptions on the potential of AI-driven analytics for patient care, as well as showing the prevalence and type of technical blocking factors that are responsible for its slow uptake. The experiments confirm that each of these issues can cause important loss of predictive model performance, depending on the model and the issue. Conclusions: Based on the qualitative study and quantitative experiments we conclude that more research on practical solutions to enable AI-driven innovation in Intensive Care Units is needed. Furthermore, the general poor situation with respect to public, usable implementations of predictive models would appear to limit the possibilities for both the scientific repeatability of the underlying research and the transfer of this research into practice.

Positioning and navigation are essential components of neuroimaging as they improve the quality and reliability of data acquisition, leading to advances in diagnosis, treatment outcomes, and fundamental understanding of the brain. Functional ultrasound (fUS) imaging is an emerging technology providing high-resolution images of the brain vasculature, allowing for the monitoring of brain activity. However, as the technology is relatively new, there is no standardized tool for inferring the position in the brain from the vascular images. This study presents a deep learning-based framework designed to address this challenge. Our approach uses an image classification task coupled with a regression on the resulting probabilities to determine the position of a single image. We conducted experiments using a dataset of 51 rat brain scans to evaluate its performance. The training positions were extracted at intervals of 375 µm, resulting in a positioning error of 176 µm. Further GradCAM analysis revealed that the predictions were primarily driven by subcortical vascular structures. Finally, we assessed the robustness of our method in a cortical stroke where the brain vasculature is severely impaired. Remarkably, no specific increase in the number of misclassifications was observed, confirming the method’s reliability in challenging conditions. Overall, our framework provides accurate and flexible positioning, not relying on a pre-registered reference but on conserved vascular patterns.

We present elPrep 5, which updates the elPrep framework for processing sequencing alignment/map files with variant calling. elPrep 5 can now execute the full pipeline described by the GATK Best Practices for variant calling, which consists of PCR and optical duplicate marking, sorting by coordinate order, base quality score recalibration, and variant calling using the haplotype caller algorithm. elPrep 5 produces identical BAM and VCF output as GATK4 while significantly reducing the runtime by parallelizing and merging the execution of the pipeline steps. Our benchmarks show that elPrep 5 speeds up the runtime of the variant calling pipeline by a factor 8-16x on both whole-exome and whole-genome data while using the same hardware resources as GATK4. This makes elPrep 5 a suitable drop-in replacement for GATK4 when faster execution times are needed.

We present elPrep 5, which updates the elPrep framework for processing sequencing alignment/map files with variant calling. elPrep 5 can now execute the full pipeline described by the GATK Best Practices for variant calling, which consists of PCR and optical duplicate marking, sorting by coordinate order, base quality score recalibration, and variant calling using the haplotype caller algorithm. elPrep 5 produces identical BAM and VCF output as GATK4 while significantly reducing the runtime by parallelizing and merging the execution of the pipeline steps. Our benchmarks show that elPrep 5 speeds up the runtime of the variant calling pipeline by a factor 8-16x on both whole-exome and whole-genome data while using the same hardware resources as GATK 4. This makes elPrep 5 a suitable drop-in replacement for GATK 4 when faster execution times are needed.

Summary: This paper describes ExaShark, a hybrid n‐dimensional array toolkit offered as a high‐level library for scientists to compute large‐scale simulations. It offers a global‐array–like interface while its runtime can be configured to use shared memory threading techniques, inter‐node distribution techniques, or combinations of both. ExaShark takes advantage of the latest HPC technologies, helping to scale to future generation systems. It has been used to develop several scientific applications including stencil codes, solvers, and matrix factorization algorithms. These applications are used to demonstrate that it improves on the state of the art by providing a user‐friendly, generic API without sacrificing performance.

Future generations of processors will exhibit an increase of faults over their lifetime, and it becomes increasingly expensive to solve the resulting reliability issues purely at the hardware level. We propose to model computations in terms of restartable task graphs in order to improve reliability at the software level. As a proof of concept, we present Cobra, a novel design for a shared-memory work-stealing scheduler that realizes this notion of restartable task graphs, and enables computations to survive hardware failures due to soft errors. A comparison with the work-stealing scheduler of Threading Building Blocks on the PARSEC benchmark suite shows that Cobra incurs no performance overhead in the absence of failures, and low performance overheads in the presence of single and multiple failures.

We repurpose a High-Throughput (cell) Imaging (HTI) screen of a glucocorticoid receptor assay to predict target protein activity in multiple other seemingly unrelated assays. In two ongoing drug discovery projects, our repurposing approach increased hit rates by 60- to 250-fold over that of the primary project assays while increasing the chemical structure diversity of the hits. Our results suggest that data from available HTI screens are a rich source of information that can be reused to empower drug discovery efforts.