The VOiCES corpus, a collaboration between Lab41 and SRI International, provides speech data recorded in acoustically challenging environments.
Lab41 outlines how Poseidon open source code uses machine learning to classify devices on a network using packet capture data.
Lab41 details their collaboration with B.Next on PySEAL, an open source Python port of the Microsoft homomorphic encryption SEAL library.
Lab41 outlines how to use source embeddings to remove dynamic noise from monaural audio recordings.
Lab41 describes when to use sampling methods, available sampling functions in TensorFlow, and how to implement custom sampling functions.
Watch Lab41's spotlight presentation at the 2017 British Machine Vision Conference in London.
Lab41 creates a data pipeline for vehicle re-identification and describes feature extraction and matching methods.
Lab41 describes chipping and feature extraction for vehicle re-identification.
Lab41 describes source-contrastive estimation to learn deep embeddings for audio source separation.
Lab41 proposes an algorithm to separate simultaneously speaking persons from each other, the “cocktail party problem”, using a single microphone.
We train a deep learning image tagging and retrieval system on large-scale user generated content using sampling methods and joint optimization of word embeddings.