archive
Every paper Pith has read. Search by title, abstract, or pith.
554 papers in cs.NE · page 12
-
Symbolic regression groups networks by generative genotype
Automatic Discovery of Families of Network Generative Processes
-
Empathic DQN cuts collateral harms by swapping agent positions
Towards Empathic Deep Q-Learning
-
Deep learning models beat others at river flow prediction
Water Preservation in Soan River Basin using Deep Learning Techniques
-
EEG signals enable machine learning to classify autism objectively
Electroencephalogram (EEG) for Delineating Objective Measure of Autism Spectrum Disorder (ASD) (Extended Version)
-
Homeostasis triggers hallucinations of learned patterns in skin model
Tactile Hallucinations on Artificial Skin Induced by Homeostasis in a Deep Boltzmann Machine
-
Reverse quantum annealing helps genetic algorithms find global optima
Quantum-Assisted Genetic Algorithm
-
Fractional-order descent improves neural net global search
Fractional-order Backpropagation Neural Networks: Modified Fractional-order Steepest Descent Method for Family of Backpropagation Neural Networks
-
Distributional successor features enable RL from noisy partial observations
A neurally plausible model learns successor representations in partially observable environments
-
VAE predicts future music values to compose new pieces
Classical Music Prediction and Composition by means of Variational Autoencoders
-
Drift analysis bounds (1+1) EA runtime on OneMax within log n
Sharp Bounds on the Runtime of the (1+1) EA via Drift Analysis and Analytic Combinatorial Tools
-
Aligned MDS tracks evolving object geometries in IT cortex
Visualizing Representational Dynamics with Multidimensional Scaling Alignment
-
Tool brings coverage testing to LSTM networks
testRNN: Coverage-guided Testing on Recurrent Neural Networks
-
Fréchet Video Distance aligns with human ratings of generated clips
Towards Accurate Generative Models of Video: A New Metric & Challenges
-
Progressive growth stabilizes GANs for 1024x1024 images
Progressive Growing of GANs for Improved Quality, Stability, and Variation
-
Search finds Swish activation that beats ReLU on deep models
Searching for Activation Functions
-
Adversarial training via projected gradient descent on the inner maximization problem…
Towards Deep Learning Models Resistant to Adversarial Attacks
-
137-billion-parameter layer beats prior best LM at 6% the compute
Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
-
Softmax peak flags neural net errors and unfamiliar inputs
A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
-
Warm restarts improve SGD anytime performance on deep nets
SGDR: Stochastic Gradient Descent with Warm Restarts
-
Real NVP gives exact likelihood and sampling for image densities
Density estimation using Real NVP
-
Wider shallower residual networks beat deep thin ones
Wide Residual Networks
-
Neural net maps camera pixels straight to steering
End to End Learning for Self-Driving Cars
-
RNNs learn to halt after variable steps
Adaptive Computation Time for Recurrent Neural Networks
-
RNNs beat item-to-item for short-session recommendations
Session-based Recommendations with Recurrent Neural Networks
-
Pipeline shrinks neural nets 35x to 49x without accuracy loss
Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
-
Student nets beat teachers with 10x fewer parameters using hints
FitNets: Hints for Thin Deep Nets
-
Gated units beat tanh RNNs on music and speech sequences
Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
-
Neural nets gain external memory and learn to copy
Neural Turing Machines
-
Attention model matches phrase-based translation quality
Neural Machine Translation by Jointly Learning to Align and Translate
-
RNN probabilities improve statistical machine translation
Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
-
Tiny changes fool neural networks and transfer across models
Intriguing properties of neural networks