Publications

A deep residual convolutional neural network for automatic lithological facies identification in Brazilian pre-salt oilfield wellbore image logs.

Estimation of permeability and effective porosity logs using deep autoencoders in borehole image logs from the Brazilian pre-salt carbonate.

Bayesian deep networks for absolute permeability and porosity uncertainty prediction from image borehole logs from Brazilian carbonate reservoirs.

In-pixel AI for lossy data compression at source for X-ray detectors.

On a method for Rock Classification using Textural Features and Genetic Optimization.

A cryogenic readout ic with 100 ksps in-pixel adc for skipper ccd-in-cmos sensors.

Texture classification based on spectral analysis and haralick features.

Rock Texture Classification Using Spectral Analysis And Genetically Optimized Texture Features.

Toward Reconfigurable In-Pixel Computing: A Fault-Tolerant Design Flow for Machine Learning Accelerators.

Large-scale imaging surveys will increase the number of galaxy-scale strong lensing candidates by maybe three orders of magnitudes beyond the number known today. Finding these rare objects will require picking them out of at least tens of millions of images, and deriving scientific results from them will require quantifying the efficiency and bias of any search method. To achieve these objectives automated methods must be developed. Because gravitational lenses are rare objects, reducing false positives will be particularly important. We present a description and results of an open gravitational lens finding challenge. Participants were asked to classify 100 000 candidate objects as to whether they were gravitational lenses or not with the goal of developing better automated methods for finding lenses in large data sets. A variety of methods were used including visual inspection, arc and ring finders, support vector machines (SVM) and convolutional neural networks (CNN). We find that many of the methods will be easily fast enough to analyse the anticipated data flow. In test data, several methods are able to identify upwards of half the lenses after applying some thresholds on the lens characteristics such as lensed image brightness, size or contrast with the lens galaxy without making a single false-positive identification. This is significantly better than direct inspection by humans was able to do. Having multi-band, ground based data is found to be better for this purpose than single-band space based data with lower noise and higher resolution, suggesting that multi-colour data is crucial. Multi-band space based data will be superior to ground based data. The most difficult challenge for a lens finder is differentiating between rare, irregular and ring-like face-on galaxies and true gravitational lenses. The degree to which the efficiency and biases of lens finders can be quantified largely depends on the realism of the simulated data on which the finders are trained.

Automatic detection of fractures and breakouts patterns in acoustic borehole image logs using fast-region convolutional neural networks.

A reconfigurable neural network ASIC for detector front-end data compression at the HL-LHC.

ABSTRACT Strong lensing is a powerful probe of the matter distribution in galaxies and clusters and a relevant tool for cosmography. Analyses of strong gravitational lenses with deep learning have become a popular approach due to these astronomical objects’ rarity and image complexity. Next-generation surveys will provide more opportunities to derive science from these objects and an increasing data volume to be analysed. However, finding strong lenses is challenging, as their number densities are orders of magnitude below those of galaxies. Therefore, specific strong lensing search algorithms are required to discover the highest number of systems possible with high purity and low false alarm rate. The need for better algorithms has prompted the development of an open community data science competition named strong gravitational lensing challenge (SGLC). This work presents the deep learning strategies and methodology used to design the highest scoring algorithm in the second SGLC (II SGLC). We discuss the approach used for this data set, the choice of a suitable architecture, particularly the use of a network with two branches to work with images in different resolutions, and its optimization. We also discuss the detectability limit, the lessons learned, and prospects for defining a tailor-made architecture in a survey in contrast to a general one. Finally, we release the models and discuss the best choice to easily adapt the model to a data set representing a survey with a different instrument. This work helps to take a step towards efficient, adaptable, and accurate analyses of strong lenses with deep learning frameworks.

We present hls4ml , a free and open-source platform that translates machine learning (ML) models from modern deep learning frameworks into high-level synthesis (HLS) code that can be integrated into full designs for field-programmable gate arrays (FPGAs) or application-specific integrated circuits (ASICs). With its flexible and modular design, hls4ml supports a large number of deep learning frameworks and can target HLS compilers from several vendors, including Vitis HLS, Intel oneAPI and Catapult HLS. Together with a wider eco-system for software-hardware co-design, hls4ml has enabled the acceleration of ML inference in a wide range of commercial and scientific applications where low latency, resource usage, and power consumption are critical. In this paper, we describe the structure and functionality of the hls4ml platform. The overarching design considerations for the generated HLS code are discussed, together with selected performance results.

Smart Pixels: In-pixel AI for on-sensor data filtering.

Ultrasound image acquisition as a didactic tool.

An Ultra-Low-Voltage, 16-Channel Current DAC ASIC in 22nm FDSOI for cryogenic SQUID Biasing.

Neural network accelerator for universal quantum control.

Porous medium permeability estimation for well imagery and characterization using complex resistivity spectra.