interpretability

An image from the COCO datset with multiple people, and the corresponding keypoint predictions.

Compressed Sensing

Intuitions with minimal working example in Python

NOV 13, 2022
categories: EXPLORATIONS
tags: COMPRESSED SENSING, COMPUTER VISION, CONVEX OPTIMIZATION, INTERPRETABILITY, LINEAR PROGRAMMING, SIGNAL PROCESSING, SPARSITY

An image is blurred out via a convolution, and then recovered via Wiener deconvolution. Notice the horizontal artifacts in the recovery, likely due to mismatching assumptions.

Understanding Wiener Deconvolution

Direct derivation, interpretation of assumptions, characterization of error bounds and code with examples

DEC 28, 2021
categories: PROOFS
tags: ALGEBRA, CALCULUS, COMPUTER VISION, CONVEX OPTIMIZATION, INTERPRETABILITY, PROOF, SIGNAL PROCESSING, SPARSITY

Part of an UMAP projection from the DCASE 2021 Task 2 dataset.

Using UMAP to Inspect Audio Data for Unsupervised Anomaly Detection under Domain-Shift Conditions

Paper presented at the DCASE 2021 Workshop

AUG 20, 2021
categories: PUBLICATIONS
tags: DCASE, DOMAIN SHIFT, INTERPRETABILITY, UMAP, UNSUPERVISED ANOMALY DETECTION

Mathematical derivation of the variational evidence lower bound.

Deriving the Variational ELBO from Bayes’ Theorem

Succint, direct and comprehensive derivation with interpretation

AUG 10, 2021
categories: PROOFS
tags: INTERPRETABILITY, MACHINE LEARNING, PROBABILITY, PROOF, VAE