Paper proposes Quaternion Approximate Networks for vision tasks
The authors report better accuracy with fewer parameters than prior convolution and quaternion models on standard vision benchmarks.
Stories per day · 30 days
Relatedadversarial-ml1arxiv1backdoor1cold-boot1computer-vision1dnn1edge-ai1measurement1model-confidentiality1model-theft1object-detection1physical-capture1quaternion-networks1robotics1stealthiness1
5 stories, newest first · one card per story
Stories tagged #image-classification
When Rotation Bounds Nothing: Streaming Image Classifiers to Field Nodes Under Physical Capture
Rotating streamed image classifiers fails to stop model recovery after physical seizure.
Quiet Triggers, Loud Footprints: A Tri-Space Measurement Study of Backdoor Stealthiness
A study finds image-classifier backdoors that look stealthy in one space often leave clear footprints in another.
Planning to Learn
Horizon loss, a one-line cross-entropy modification, improves top-1 accuracy on ImageNet and MNIST, with gains increasing under label noise.
ALPINE: Adaptive Localization for Parameter- and Sample-Efficient Few-Shot Learning
ALPINE is an ultra-lightweight few-shot image classifier that beats Prototypical Networks, Relation Networks, and MAML with 27-53% fewer parameters.