Activation functions have a notorious impact on neural networks on both training and testing the models against the desired problem. Currently, the most used activation function is the Rectified Linear Unit (ReLU). This paper introduces a new and novel activation function, closely related with the new activation $Swish…
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
169,051 papers · 148 categories
Trend · papers per month
25 results for “Swish”
This work interprets GELU and related activations via a first-order loss function.
problem Understanding and optimizing activation functions in neural networks.
method Complementary interpretation using the Gaussian first-order loss function.
result Calibrated or learned uniform-threshold gates are competitive and often outperform GELU, ReLU, and SiLU/Swish.
LSQ+ improves quantization of neural nets with Swish activations, achieving state-of-the-art results.
problem Quantization of neural nets with Swish activations, especially negative activations, leads to significant performance loss.
method Introduces learnable scale and offset parameters for asymmetric quantization, and uses MSE-based initialization for quantization parameters.
result Significantly outperforms LSQ for low-bit quantization of neural nets with Swish activations, achieving up to 5.6% gain with W2A2 quantization of EfficientNet-B0.
A new activation function FTS improves deep learning performance.
problem Hindered propagation of negative values in ReLU.
method Proposed Flatten-T Swish (FTS) activation function, evaluated on MNIST dataset.
result FTS with T=-0.20 improves MNIST classification accuracy by 1.15% on 8-layer DFNN.
We present a Statistical Mechanics (SM) model of deep neural networks, connecting the energy-based and the feed forward networks (FFN) approach. We infer that FFN can be understood as performing three basic steps: encoding, representation validation and propagation. From the meanfield solution of the model, we obtain a…
The weight initialization and the activation function of deep neural networks have a crucial impact on the performance of the training procedure. An inappropriate selection can lead to the loss of information of the input during forward propagation and the exponential vanishing/exploding of gradients during back-propag…
Highly Accurate CNN Inference Using Approximate Activation Functions over Homomorphic Encryptioncs.LG
Improved CNN accuracy for encrypted data using approximate activation functions.
problem Low accuracy in classifying encrypted data using homomorphic encryption.
method Used a fourth-order polynomial approximation of the Swish activation function with batch normalization for homomorphic encryption.
result Achieved 99.22% accuracy on MNIST and 80.48% on CIFAR-10, improving by 0.04% and 4.11% respectively.
Study uses 1D-CNNs to forecast mortality in ELSA survey.
problem Forecasting mortality in the middle-aged and older population.
method 1D-CNNs applied to longitudinal data with various over/undersampling and activation functions.
result Swish nonlinearity outperforms other functions in forecasting mortality.
Study learns a neuron with non-monotonic activation functions.
problem Learning a single neuron with non-monotonic activation functions.
method Gradient descent (GD) with conditions on activation function and input distribution.
result Learnability of non-monotonic activation functions is established without monotonicity assumption.
This paper proposes an embedding-based neural network for more accurate investment return prediction.
problem Accurately predicting investment returns requires understanding industry knowledge and news, as well as leveraging relevant theories.
method The approach uses embedding to encode investment IDs into low-dimensional vectors, leveraging dual branches to separate different information, and employs the swish activation function.
result The proposed embedding-based dual branch model outperforms traditional machine learning models like Xgboost, Lightgbm, and Catboost on the Ubiquant Market Prediction dataset.
Deep networks can approximate various activation functions with modest adjustments.
problem Expressive power of deep neural networks with diverse activation functions.
method Approximation of any activation function in set A by ReLU networks with specific scaling factors.
result Approximation of any activation function in a specific subset of A by ReLU networks with (1,1) scaling factors.
This paper compares different activation functions in GLVQ models.
problem Improving performance of GLVQ models using different activation functions.
method Investigates and compares ReLU, sigmoid, and swish activation functions in GLVQ models.
result Different activation functions have varying impacts on GLVQ model performance.
Gradient descent with logistic loss can interpolate deep networks with smoothed ReLU activations under certain conditions.
problem Conditions for gradient descent to drive logistic loss to zero in deep networks with smoothed ReLU activations.
method Gradient descent applied to fixed-width deep networks with smoothed ReLU approximations (e.g., Swish, Huberized ReLU).
result Gradient descent can drive logistic loss to zero under specific conditions, providing bounds on convergence rate.
Paper extends understanding of deep network nonlinearities using vector quantization and statistical inference.
problem Limited understanding of deep network nonlinearities, especially non-piecewise affine and non-convex functions.
method Link deterministic max-affine spline operators to probabilistic Gaussian Mixture Models (GMMs) for a broader class of nonlinearities.
result Enforces orthogonality in linear filters can significantly improve deep network performance.
New method learns neural network activation functions from data.
problem Learning activation functions for neural networks.
method Model each neuron's activation function as a small neural network.
result Learned activation functions improve network performance.
New method improves ResNet performance without batch normalization.
problem Improving ResNet performance without batch normalization.
method Adapted Weight Standardization to maintain signal propagation.
result Highly performant ResNets achieve state-of-the-art performance on ImageNet.
K-TanH is a fast, efficient deep learning activation function.
problem Improving the speed and efficiency of deep learning models.
method K-TanH is a novel, highly accurate, low-precision integer approximation of TanH.
result K-TanH achieves up to 5x speedup over Intel SVML and is more efficient.
Mish is a new activation function that improves neural network performance.
problem Improving the performance and training dynamics of neural networks.
method Mish is a self-regularized non-monotonic activation function defined as . It outperforms other functions on benchmarks like ImageNet-1k and MS-COCO.
result Mish outperforms Leaky ReLU and ReLU on benchmarks like MS-COCO and ImageNet-1k, respectively, with comparable network parameters.
Recently, self-normalizing neural networks (SNNs) have been proposed with the intention to avoid batch or weight normalization. The key step in SNNs is to properly scale the exponential linear unit (referred to as SELU) to inherently incorporate normalization based on central limit theory. SELU is a monotonically incre…
Geometric theory of projection heads in self-supervised learning.
problem Dimensional collapse and information invariance trade-off in projection heads.
method Geometric modeling of projection heads as Riemannian metrics, analyzing Hessian eigenvalues, and tracking optimization geometry.
result Smooth nonlinear heads induce negative curvature, preventing collapse; linear and ReLU heads cannot.
Single neuron with ADA learns XOR and outperforms other functions.
problem Classifying linearly non-separable data.
method Proposed a new artificial neuron with apical dendrite activation.
result ADA function achieves 100% accuracy on XOR and superior performance on benchmark datasets.
This paper precisely estimates transformer derivatives for explicit learning guarantees.
problem Computing fully-explicit generalization bounds for transformers with precise higher-order derivative estimates.
method Analyzes and estimates all higher-order derivatives of transformers with multiple attention heads and layer normalization.
result Obtains explicit pathwise generalization bounds for transformers learning from non-i.i.d. samples.
This paper explores adversarial training limits and improves model robustness against norm-bounded perturbations.
problem Understanding and improving adversarial robustness of deep neural networks.
method Systematic study of adversarial training with various factors, including model size, activation functions, and unlabeled data.
result Training robust models that go beyond state-of-the-art results by combining larger models, Swish/SiLU activations, and model weight averaging.
New self-exciting random evolutions (SEREs) for modeling traffic and transport processes.
problem Modeling self-exciting and clustering effects in traffic and transport processes.
method Introducing a new process based on a superposition of a Markov chain and a Hawkes process, and constructing self-exciting random evolutions (SEREs).
result Developed new models and limit theorems for SEREs, including averaging and diffusion approximation.
The study examines how different activation functions affect the training of overparametrized neural networks.
problem Understanding the impact of activation functions on the training of overparametrized neural networks.
method Theoretical analysis of 2-layer neural networks with various activation functions.
result The performance of non-smooth activations like ReLU, SELU, and ELU is robust under minimal assumptions, while smooth activations like tanh and polynomials can have varying eigenvalues.