neural network architectures
Neural network architectures refer to the specific design and arrangement of layers and nodes in a neural network, affecting how data is processed and learned, including variations like convolutional networks, recurrent networks, and transformers.
- Compositional Reasoning with Transformers, RNNs, and Chain of Thought
- Differentiation Through Black-Box Quadratic Programming Solvers
- Flow Equivariant Recurrent Neural Networks
- Hadamax Encoding: Elevating Performance in Model-Free Atari
- Improving Deep Learning for Accelerated MRI With Data Filtering
- Influence Functions for Edge Edits in Non-Convex Graph Neural Networks
- Non-Singularity of the Gradient Descent Map for Neural Networks with Piecewise Analytic Activations