How DeepRRP Works
summary
Company deepkapha.ai explains a neuroscience-led research stack, a new activation family called ARiA (Adaptive Richards-curve weighted activation) derived from Richards growth curves, and concrete benchmarks: ARiA variants (ARiA1/ARiA2) plug into DenseNet, WideResNet and VGG16, use GPU/TPU training, and show ~3% accuracy gain on CIFAR-10/100 and ~6% gain on glioma classification versus ReLU/Swish. Also pursuing thalamus-to-cortex modeling to inform capsule‑style architectures, building a data-curation product (Ellie) and medical prediction pipelines with published field trials.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 5:17Deep Kapha mission
- 28:17Research drives deep learning
- 41:52Neuroscience insights
- 55:13Aria activation function
- 1:16:15Deep Kapha products
- 1:30:14Q&A
related talks
37mDeep Recommender Systems
29mDeep Learning, Neuroscience and the future of AI
47mInterpreting Deep Neural Networks (DNNs)
21mAdvanced Ray RLlib: New Features and New Possibilities
26mAnkur Narang; Senior Vice President – Technology & Decision Sciences at Yatra at Cypher2017- Part1/3
28mAI in healthcare - Deep learning for precision psychiatry
37mWhat's new in Deep Learning?
13mBuilding High-Performance AI Literature Processing