Page Not Found
Page not found. Your pixels are in another canvas.
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Page not found. Your pixels are in another canvas.
About me
This is a page not in th emain menu
Published:
Jupyter RISE slideshow presentation motivating and evaluating the use of deep learning inference-as-a-service in gravitational wave use cases.
Published:
Overview of tools to accelerate inference for deep recommendation models on GPU.
A modular, production-ready framework for training neural networks to detect binary blackhole merge signals in the presence of non-Gaussian background noise, with a focus on efficiency and scale.
A framework similar to BBHNet for using neural networks to regress from witnessed environmental noise to noise observed in the interferometer strain channel to increase search sensitivity.
hermes
PermalinkTools for accelerating, serving, and making streaming asynchronous inference requests to complex ensembles of deep learning models at scale.
ml4gw
PermalinkTorch utilities for training neural networks for gravitational wave physics applications.
High level introduction to ML concepts starting from rules-based AI and introducing ML as an extension of it. Working on a follow-up to introduce the concepts of over-fitting and complexity. Good example of how to leverage Github Workflows for publishing.
Pipeline for training ResNet50 using TensorFlow in a Kaggle notebook, with hypeparameter search performed by cross-validation with HyperBand.
Now out-of-date project using the Tensorflow Estimator API to asynchronously train multiple MLP workers on sparse data on a single GPU. Planning to update to more modern Keras syntax when I have the time.
Analyzing airport arrival lead time as a cost-optimization problem, with solutions estimated from suspect TSA data.
AWS Elastic Beanstalk/RDS application for a DIY fantasy surf league.
Published in arXiv CoRR, 2017
Modelling patient physiological state as a dynamical system in order to forecast mortality outcomes
Download here
Published in arXiv GR-QC, 2021
Presenting advantages in throughput and latency wrought by deep learning inference-as-a-service in both online and offline gravitational wave analysis, as well as outlining a paradigm for further optimization.
Download here
Published in FlexScience 2022, 2022
Outlining the deployment requirements of a particular machine learning model for use in production gravitational wave physics pipelines and describing a set of tools for simplifying how they can be achieved.
Download here
Published:
Demo on converting trained recurrent and convolutional models from multiple deep learning libraries into accelerated inference runtime engines and exposing them via an inference service
Published:
Demonstrating a simple workflow for building embedded applications using deep learning models trained in the cloud
Published:
Discussing a simple end-to-end workflow for leveraging Tensorflow APIs to preprocess audio classification data, train a deep learning model on it, then serve it for inference at reduced precision
Published:
Discussing an accelerated inference deployment of a common recommendation architecture using a custom embedding kernel with TensorFlow APIs.
Published:
In gravitational-wave detectors, regression techniques are applied to remove noise artifacts in order to improve the ability to observe and extract information from astrophysics signals. We present a deep learning-based noise regression method called DeepClean that can subtract linear and non-linear noise in gravitational-wave data from the Advanced LIGO detectors. We also discuss our work toward a new computing model in gravitational-wave data analysis where GPU and FPGA acceleration on machine learning inference can be deployed on an as-a-service basis. We use DeepClean as a use-case for exploring such computing models in order to achieve real-time capabilities and overall flexibility such models provide.