6 papers
No time to train! Training-Free Reference-Based Instance Segmentation
Miguel Espinosa, Chenhongyi Yang, Linus Ericsson +2
The performance of image segmentation models has historically been constrained by the high cost of collecting large-scale annotated data. The Segment Anything Model (SAM) alleviate…
Evolutionary Architecture Search through Grammar-Based Sequence Alignment
Adri Gómez MartÃn, Felix Möller, Steven McDonagh +5
Neural architecture search (NAS) in expressive search spaces is a computationally hard problem, but it also holds the potential to automatically discover completely novel and perfo…
ONNX-Net: Towards Universal Representations and Instant Performance Prediction for Neural Architectures
Shiwen Qin, Alexander Auras, Shay B. Cohen +4
Neural architecture search (NAS) automates the design process of high-performing architectures, but remains bottlenecked by expensive performance evaluation. Most existing studies…
Transferrable Surrogates in Expressive Neural Architecture Search Spaces
Shiwen Qin, Gabriela Kadlecová, Martin Pilát +5
Neural architecture search (NAS) faces a challenge in balancing the exploration of expressive, broad search spaces that enable architectural innovation with the need for efficient…
Hyperparameter Selection in Continual Learning
Thomas L. Lee, Sigrid Passano Hellan, Linus Ericsson +2
In continual learning (CL) -- where a learner trains on a stream of data -- standard hyperparameter optimisation (HPO) cannot be applied, as a learner does not have access to all o…
There is no SAMantics! Exploring SAM as a Backbone for Visual Understanding Tasks
Miguel Espinosa, Chenhongyi Yang, Linus Ericsson +2
The Segment Anything Model (SAM) was originally designed for label-agnostic mask generation. Does this model also possess inherent semantic understanding, of value to broader visua…