If you are in Taipei for #asru2023 , Come check out our Poster.

In a new paper we present a simple projection module to facilitate the use of synthetic speech to improve personalisation in ASR models. Amazon Science. #interspeech2023 . amazon.science/publications/e…

Glad to share our paper 'Using model explanations to guide deep learning models towards consistent explanations for EHR data' at Nature Scientific Reports, by Matthew Watson Comp Sci @ Durham with Bashar Awwad. Big 🙏 to Evergreen Life for sponsoring our work💚nature.com/articles/s4159…

We have liftoff! Good luck to NASA, Lockheed Martin and Cisco with the Artemis 1 mission, which will send Alexa into space for the first time to explore how voice tech and AI can assist future astronauts. Learn more about the mission. #Artemis

Very glad to see our work Comp Sci @ Durham with Bashar Awwad being featured in RSIP Vision.
rsipvision.com/ComputerVision…

A peek into some of the challenges we faced while developing Alexa for Arabic. amazon.science/latest-news/ho…

Great result for my collaboration with Noura Al Moubayed and Matthew Watson Comp Sci @ Durham

And Launch 🚀Alexa can now speak Arabic! arabnews.com/node/1982586/b…

We need your help to support STEM in schools.

Personal update. I am excited to share a new direction in my career. I am leaving the FinTech field and starting a new position with the research team at Amazon Alexa.

That is good enough reason to upgrade to 3.10.

How much confidence you have in the explanation of your model? We demonstrate in this paper the variability in model explanation under different training settings.

One of the best books I read on machine learning.

We are looking for a senior data scientist to join our team. Get in touch and be part of the team changing the landscape of anti-money laundering investigations globally.
linkedin.com/jobs/view/2470…

This is really coocerning. It does highlight issues in both data pre-processing and model building. There is a need for a strong regulatory framework to ensure trust and transparency in deployed models.

Integrating experts' knowledge with machine learning models is quite a challenge. Chris Brannigan talks about our approach in anti money laundering.

If reviewer 3 is behaving like a #reviewer2 , do i have to ask the area chair to relabel them? #CVPR2021

3 days of food for 1 family...
Just not good enough.

This is really coocerning. This can only increase the risk of people not taking up the vaccine. Another reckless step by incompetent government. bbc.co.uk/news/uk-555037…

This is beautifully scary!