Got a new preprint on arXiv, which introduces HARD-Training. When classifying depression severity from speech, we demonstrate there can be an increase in classifier performance when using just the training samples nearest the decision boundary: arxiv.org/abs/2206.01542

Nick Cummins @EdwardC22936056 Jude Dineley Dr Faith Matcham Femke Lamers RADAR-CNS

Having a great time at the ISCA-SAC round table!

It was a huge team effort, thanks to all my co-authors for their assistance, this includes everyone tagged in the main tweet, and Dr Grace Lavelle, @_Dr_Daniel, Katie White, Carolin Oetzmann, Edward L. Campbell,Sara Simblett, Yatharth Ranjan, Zulqarnain Rashid, Raquel Bailón Luesma, Björn Schuller

Don't forget - today June 1st is the deadline to get the camera-ready version of your paper uploaded. Each paper must have one author registered so that we know your paper is covered. The programme will be out by June 20th. Dublin ahoy! ☘️