Learning more about manifold learning algorithms has been on my TODO list for a while. I also wanted to play with Captum for a while as well. The feature_analysis project started with the desire to dive into the details of t-SNE and UMAP and the willingness to dive further into the neural network feature, specifically convolutions, and then onto feature-based searches.
Read More / Live Demo! / Source CodeSuneeta Mall
Rambling of a curious engineer & data scientist
Errors in datasets are unavoidable - not just because Murphy’s Law always holds but because it’s a complicated process involving humans, perception, cognition, and software and systems! So, how do we manage them? It’s something that I have been thinking a lot about lately. I, recently came across “Confidant learning”, a proposed technique by Curtis Northcutt and his group. It piqued my interested so I spent sometime exploring Cleanlab.
Read More / Live Demo! / Source CodeVery excited to share that based on the KubeCon ‘19 talk and reproducible-ML project and E2E Reproducible ML project, O’reilly reached out to create an interactive (katacoda) scenario series deep-diving into challenges and solutions in realizing reproducible ML.
Read More / Live Demo! / Source CodeBased on Oxford Pet dataset, standard semantic segmentation example by tensorflow is extended to demonstrate how end to end reproducible machine learning can be realized e2e-ml-on-k8s.
Read More / Live Demo! / Source CodeKubernetes Continous Delivery (kcd), formally known as Container Version Manager (cvmanager), is a continous integration (CI) and continous delivery (CD) tool designed for Kubernetes cluster/services. Fundamentally, kcd is a custom Kubernetes controller to achieve a declarative configuration approach to continuous deployment.
Read More / Live Demo! / Source CodeGoReportCard Lite, a slim version of original Go Report Card by Shawn Smith and Herman Schaaf catered to be integrated into CI pipelines and can generate reports using local/private repositories.
Read More / Live Demo! / Source CodeGatling, is a powerful tool for performance and load testing. There is no easy to create and scale out Gatling based load. Gatling on ECS was written with intention to scale out load test using AWS elastic container services. For more information see project.
Read More / Live Demo! / Source Code