Efficient nearest neighbors inspired by the fruit fly brain

The New algorithm

The Biological analogy

Procedure for hashing the smell in a fruit fly’s brain

Efficiency of generating the hash

Implementation and Results

It works as advertised! I got 3~4x higher mAP scores from Fly LSH for the same computational budget.

Can we do better?

Learning the directions along which to take projections improves the performance of Fly LSH
For hash length k=32 (a) Directions assigned randomly, and (b) Directions learnt by the autoencoder. Both images are 784x640 matrices. In (b) the highest 10% weights along each column were binarized to 1.
An autoencoder with sparse weights performs slightly worse on MNIST. I appreciate any ideas to improve this.

TL;DR

--

--

--

Blogs about replicating research papers in machine learning

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Real-Time Dense Stereo Matching with ELAS on FPGA Accelerated Embedded Devices

Dog Breed Identification using Deep Learning

Emgu CV : Setting up Logistic Regression

Covariate Shift in Malware Classification

How to Complete the DLND Project 4 (TV Script Generation) in Nine Steps

深度學習筆記(16):Convolution Neutral Network

SIIM FISABIO RSNA COVID-19 DETECTION

Improving performance of a neural network

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Jaiyam Sharma

Jaiyam Sharma

Blogs about replicating research papers in machine learning

More from Medium

Overfitting: A headache you want to do away with.

Overfitting

Achieving True Accuracy Using Statistics

Machine learning Lecture

The future impact of man-made intellectual ability