Michael E. Davies: Inexact gradient projection and fast data driven comp... youtu.be/-2-M6NMud2c via YouTube

Three good reasons to use black-box AI systems: science.sciencemag.org/content/364/64…

“researchers’ careers depend more on publishing results with ‘impact’ than on publishing results that are correct” This is not a revelation but it is still shocking. You can read the full nature article about research misconduct at Duke University here nature.com/articles/d4158…

Oldies but goldies: G. Cybenko, Approximations by superpositions of sigmoidal functions, 1989. Proves that a 1 hidden layer perceptron can approximate arbitrary continuous functions using a large enough number of neutrons. en.wikipedia.org/wiki/Universal… en.wikipedia.org/wiki/Multilaye…

We are recruiting! Research Associate in Image Processing & Machine Learning for Microendoscopy: An exciting opportunity to develop algorithms for a breakthrough healthcare imaging system vacancies.ed.ac.uk/pls/corehrrecr…. Healthcare Impact Partnership (EP/S025987/1)." EPSRC Proteus IRC

Fully funded PhD still available in my group! Great opportunity to develop new computational methods for imaging through turbulence.Heriot-Watt Research@EPS HWU SIPLab Edinburgh Leonardo in the UK 🇬🇧 CENSIS ISSS Heriot-Watt

Seeing a lot of confusion regarding these two terms in GAN papers lately. Mode Collapse: When a large region of a model's input space maps to a small region around a single (often bad) sample. Mode Dropping: When modes in the data are not represented in the output of the model.

In the first of our Winter Graduation Stories 2019, MSc Signal Processing and Communications student Xiao Liu explains why he chose to study at The University of Edinburgh, picks his highlights of being a student and explains what he plans to do next. edin.ac/37uG9kr #EdinburghGrad✨🎓

This is our recent collaborative work, between Engineering and Chemistry in Edinburgh, with the support of industry. Looking forward to extending to other new- phase/material detection. Stephanie Brookes @ColinCa57082954

Which dataset would you choose for training the reconstruction net of an imaging system? 1. Pairs of images+measurements 2. Just all raw measurements Spoiler alert: the obvious answer is incorrect - a thread 🧵

Please re-tweet: PhD position in causality, representation learning & applications in healthcare. Fully funded international fees too (by Canon Medical Research Europe Ltd. Royal Academy of Engineering). Co-supervised by Alison Q O'Neil. Join Pedro P. Sanchez @Xiaoliu33745946 and others in VIOS. findaphd.com/phds/project/c…

A great opportunity to do a PhD in an excellent team.

Survival Strategies for Depressed AI Academics: arxiv.org/abs/2304.06035

Next talk by Mehrdad Yaghoobi on “self-supervised hyperspectral inpainting with the optimisation inspired deep neural network” at #iscs23

📢📢 Release of DeepInverse library 📢📢 After months of intense work, we are releasing the first stable version of DeepInverse deepinv.github.io/deepinv/, a PyTorch library for solving inverse problems with deep learning. with Samuel Hurault, Matthieu Terris and Dongdong Chen A 🧵

📢📢📢 Job Opportunity! Join Dr Elliot J. Crowley and the Bayesian and Neural Systems Group as a postdoc in NAS and AutoML at the University of Edinburgh. 24-month funded position starting Autumn 2023. PhD in ML or related field required. Apply now: elxw.fa.em3.oraclecloud.com/hcmUI/Candidat…

An excellent opportunity in a world leading group.

Interesting stuff, spectrum.ieee.org/view-from-the-…