archive
Every paper Pith has read. Search by title, abstract, or pith.
554 papers in cs.NE · page 11
-
Genetic algorithms fold human suggestions into starshade schedules
Genetic Algorithms for Starshade Retargeting in Space-Based Telescopes
-
Neural filter cuts analog layout simulations by 100x
BagNet: Berkeley Analog Generator with Layout Optimizer Boosted with Deep Neural Networks
-
Dynamic inputs let surrogates replace many RL fitness evaluations
Surrogate Models for Enhancing the Efficiency of Neuroevolution in Reinforcement Learning
-
Quality diversity tunes zebrafish models more accurately than standard evolution
Automatic Calibration of Artificial Neural Networks for Zebrafish Collective Behaviours using a Quality Diversity Algorithm
-
Novelty sampling yields predictive supernets for NAS
Efficient Novelty-Driven Neural Architecture Search
-
First stochastic DNN framework runs on AQFP superconducting logic
A Stochastic-Computing based Deep Learning Framework using Adiabatic Quantum-Flux-Parametron SuperconductingTechnology
-
Streaming graphs and topology cubes scale analysis to millions of samples
Scalable Topological Data Analysis and Visualization for Evaluating Data-Driven Models in Scientific Applications
-
Transfer cuts NAS search from 200 to 6 GPU days
XferNAS: Transfer Neural Architecture Search
-
Random fixed weights reach SOTA reinforcement learning scores
Convolutional Reservoir Computing for World Models
-
Invertible networks generate and classify EEG signals above chance
Deep Invertible Networks for EEG-based brain-signal decoding
-
Neural network filter speeds genetic search for trailer fleets
Machine Learning based Simulation Optimisation for Trailer Management
-
Output differences embed variable neural nets for performance prediction
Prediction of neural network performance by phenotypic modeling
-
Deep networks match human numerosity judgments
Perception of visual numerosity in humans and machines
-
Shifted-ReLU often matches BN accuracy in 1-bit weight CNNs
Single-bit-per-weight deep convolutional neural networks without batch-normalization layers for embedded systems
-
Interactive QD models user picks to penalize drift
Modeling User Selection in Quality Diversity
-
Anticipation module speeds EFS recovery after abrupt drifts
ParaFIS:A new online fuzzy inference system based on parallel drift anticipation
-
Temporal differencing replaces activation derivatives in backprop
Iterative temporal differencing with random synaptic feedback weights support error backpropagation for deep learning
-
Review groups deep learning segmentation methods by contribution
Understanding Deep Learning Techniques for Image Segmentation
-
Evolvability ES maximizes mutation diversity for fast adaptation
Evolvability ES: Scalable and Direct Optimization of Evolvability
-
Intermediate routing makes capsule networks faster and more accurate
Using dynamic routing to extract intermediate features for developing scalable capsule networks
-
Linear algorithm trains million-variable RBMs competitively
An Evolutionary Algorithm of Linear complexity: Application to Training of Deep Neural Networks
-
Olfactory steps let one network handle gas
Signal Conditioning for Learning in the Wild
-
Batch-lexicase yields more generic rules in classifier systems
Lexicase selection in Learning Classifier Systems
-
Competitive networks learn patterns via Haeussler and Firefly wiring
Learning in Competitive Network with Haeusslers Equation adapted using FIREFLY algorithm
-
Spiking network sustains tracking from sparse events
Event-based attention and tracking on neuromorphic hardware
-
Robots build semantic maps and cluster places online without forgetting
Incremental Semantic Mapping with Unsupervised On-line Learning
-
Transformers capture lexical semantics better than RNNs in NMT
An Intrinsic Nearest Neighbor Analysis of Neural Machine Translation Architectures
-
Neural nets let active inference handle complex control tasks
Deep Active Inference as Variational Policy Gradients
-
Quality Diversity generates agent pools for Hanabi ad-hoc tests
Diverse Agents for Ad-Hoc Cooperation in Hanabi
-
Hybrid algorithm yields better PSP solutions on triangular lattice
A new hybrid genetic algorithm for protein structure prediction on the 2D triangular lattice
-
Heuristic selection lifts ML accuracy on smart-grid FDI attacks
Smart Grid Cyber Attacks Detection using Supervised Learning and Heuristic Feature Selection
-
Evolutionary algorithm translates Sinhala sentences to English
Evolutionary Algorithm for Sinhala to English Translation
-
Flappy Bird trained from raw pixels with A3C and DQN
Playing Flappy Bird via Asynchronous Advantage Actor Critic Algorithm
-
Genetic algorithm evolves cells to 96% on CIFAR-10 with shared weights
Genetic Network Architecture Search
-
Four criteria organize online RNN training algorithms into performance clusters
A Unified Framework of Online Learning Algorithms for Training Recurrent Neural Networks
-
Pointer buffer replaces data copy in convolutions
The Indirect Convolution Algorithm
-
Tool-use transfers generate multiple RL test sets for distinct generalization
Reasoning and Generalization in RL: A Tool Use Perspective
-
Evolved nets reach lower regression error with fewer parameters
Neural Network Architecture Search with Differentiable Cartesian Genetic Programming for Regression
-
Cellular automata recast as neural networks for reservoir computing
A general representation of dynamical systems for reservoir computing
-
Evolutionary search balances Hearthstone decks near 50% wins
Evolving the Hearthstone Meta
-
Stack models generalize on arithmetic where Turing machines fail
Understanding Memory Modules on Learning Simple Algorithms
-
MOSFET synapses enable on-chip learning in analog neural nets
On-chip learning in a conventional silicon MOSFET based Analog Hardware Neural Network
-
Publication counts rank 14 optimizers across 18 fields from 2007-2017
A Note On The Popularity of Stochastic Optimization Algorithms in Different Fields: A Quantitative Analysis from 2007 to 2017
-
Nanowires produce neuron-like spikes efficiently
A Power Efficient Artificial Neuron Using Superconducting Nanowires
-
Pruned ResNet-50 keeps same interpretable units past 90% compression
Dissecting Pruned Neural Networks
-
Spiking neurons cut asset allocation time by 1000x at 99.9% accuracy
High Speed Cognitive Domain Ontologies for Asset Allocation Using Loihi Spiking Neurons
-
ARMIN cuts memory network overhead below LSTM levels
ARMIN: Towards a More Efficient and Light-weight Recurrent Memory Network
-
Adaptive sparse networks generalize better than dense ones
On improving deep learning generalization with adaptive sparse connectivity
-
Passive memristor crossbar classifies MNIST near software limit
4K-Memristor Analog-Grade Passive Crossbar Circuit
-
Hybrid PSO learns user-specific summary weights from labeled examples
User-Oriented Summaries Using a PSO Based Scoring Optimization Method