arXiv research

This paper presents a new artificial neuron model capable of learning its receptive field in the topological domain of inputs. The model provides adaptive and differentiable local connectivity (plasticity) applicable to any domain. It requires no other tool than the backpropagation algorithm to learn its parameters whi…

In order to classify linearly non-separable data, neurons are typically organized into multi-layer neural networks that are equipped with at least one hidden layer. Inspired by some recent discoveries in neuroscience, we propose a new model of artificial neuron along with a novel activation function enabling the learni…

Stable unactivated neurons reduce expressiveness in ReLU networks.

In this paper, a simple, general method of adding auxiliary stochastic neurons to a multi-layer perceptron is proposed. It is shown that the proposed method is a generalization of recently successful methods of dropout (Hinton et al., 2012), explicit noise injection (Vincent et al., 2010; Bishop, 1995) and semantic has…

Local convergence theory for mildly over-parameterized neural nets.

Unseen or out-of-domain data can seriously degrade the performance of a neural network model, indicating the model's failure to generalize to unseen data. Neural net pruning can not only help to reduce a model's size but can improve the model's generalization capacity as well. Pruning approaches look for low-salient ne…

Wide neural networks' last hidden layers split into groups of redundant neurons.

Training state-of-the-art, deep neural networks is computationally expensive. One way to reduce the training time is to normalize the activities of the neurons. A recently introduced technique called batch normalization uses the distribution of the summed input to a neuron over a mini-batch of training cases to compute…

3-layer NTK models generalize better than 2-layer models, especially with large input dimensions.

Binary Stochastic Filtering (BSF), the algorithm for feature selection and neuron pruning is proposed in this work. The method defines filtering layer which penalizes amount of the information involved in the training process. This information could be the input data or output of the previous layer, which directly lead…

Deep neural networks (DNNs) are known for extracting useful information from large amounts of data. However, the representations learned in DNNs are typically hard to interpret, especially in dense layers. One crucial issue of the classical DNN model such as multilayer perceptron (MLP) is that neurons in the same layer…

A measure of neural complexity quantifies how hard it is to access information across neurons.

Optimal neuron activation functions improve neural network performance.

New training method for ReLU networks achieves optimal weight size for memorization.

Gradient descent dynamics in neural networks show quenching and activation phases.

Extreme learning machine (ELM) is a new single hidden layer feedback neural network. The weights of the input layer and the biases of neurons in hidden layer are randomly generated, the weights of the output layer can be analytically determined. ELM has been achieved good results for a large number of classification ta…

Elite ONNs learn better with synaptic plasticity, improving performance over CNNs.

Statistical neurodynamics studies macroscopic behaviors of randomly connected neural networks. We consider a deep layered feedforward network where input signals are processed layer by layer. The manifold of input signals is embedded in a higher dimensional manifold of the next layer as a curved submanifold, provided t…

A new framework for flexible neural network receptive fields.

This paper introduces a hierarchical associative memory model with multiple layers.

The permutation symmetry of neurons in each layer of a deep neural network gives rise not only to multiple equivalent global minima of the loss function, but also to first-order saddle points located on the path between the global minima. In a network of $d-1$ hidden layers with $n_k$ neurons in layers $k = 1, \ldots, …

We demonstrate that a very deep ResNet with stacked modules with one neuron per hidden layer and ReLU activation functions can uniformly approximate any Lebesgue integrable function in $d$ dimensions, i.e. $\ell_1(\mathbb{R}^d)$. Because of the identity mapping inherent to ResNets, our network has alternating layers of…

The study analyzes how many neurons are needed for two-layer neural networks trained with gradient descent.

Two-layer neural networks need more neurons to be robust.

One of the main difficulties in analyzing neural networks is the non-convexity of the loss function which may have many bad local minima. In this paper, we study the landscape of neural networks for binary classification tasks. Under mild assumptions, we prove that after adding one special neuron with a skip connection…

This work makes vision networks more interpretable by identifying key neurons.

Hysteresis phenomena have been observed in different branches of physics and engineering sciences. Therefore, several models have been proposed for hysteresis simulation in different fields; however, almost neither of them can be utilized universally. In this paper by inspiring of Preisach Neural Network which was insp…

Neuron-specific dropout reduces overfitting and data needs for neural networks.

Recent advancements in language representation models such as BERT have led to a rapid improvement in numerous natural language processing tasks. However, language models usually consist of a few hundred million trainable parameters with embedding space distributed across multiple layers, thus making them challenging t…

Recently, deep learning has achieved huge successes in many important applications. In our previous studies, we proposed quadratic/second-order neurons and deep quadratic neural networks. In a quadratic neuron, the inner product of a vector of data and the corresponding weights in a conventional neuron is replaced with…

It is well-known that neural networks are universal approximators, but that deeper networks tend in practice to be more powerful than shallower ones. We shed light on this by proving that the total number of neurons $m$ required to approximate natural classes of multivariate polynomials of $n$ variables grows only line…

SGD efficiently learns the XOR function with near-optimal sample complexity.

We show that a collection of Gaussian mixture models (GMMs) in $R^{n}$ can be optimally classified using $O(n)$ neurons in a neural network with two hidden layers (deep neural network), whereas in contrast, a neural network with a single hidden layer (shallow neural network) would require at least $O(\exp(n))$ neurons …

Paper introduces supervised and unsupervised TAM models for binary neurons.

While neural networks are powerful approximators used to classify or embed data into lower dimensional spaces, they are often regarded as black boxes with uninterpretable features. Here we propose Graph Spectral Regularization for making hidden layers more interpretable without significantly impacting performance on th…

Improves conditions for mode connectivity in deep neural networks.

Despite our extensive knowledge of biophysical properties of neurons, there is no commonly accepted algorithmic theory of neuronal function. Here we explore the hypothesis that single-layer neuronal networks perform online symmetric nonnegative matrix factorization (SNMF) of the similarity matrix of the streamed data. …

New $L_1$ regularization controls neural network generalization error and sparsifies input dimensions.

New measure FTC quantifies how much a ReLU network can fine-tune.

AGF explains feature learning in neural networks through alternating steps.

The challenge of assigning importance to individual neurons in a network is of interest when interpreting deep learning models. In recent work, Dhamdhere et al. proposed Total Conductance, a "natural refinement of Integrated Gradients" for attributing importance to internal neurons. Unfortunately, the authors found tha…

Self-ONNs adapt nodal operators during training for higher diversity and efficiency.

CHANI learns classification tasks with local transformations inspired by biology.

Recent results in nonparametric regression show that deep learning, i.e., neural network estimates with many hidden layers, are able to circumvent the so-called curse of dimensionality in case that suitable restrictions on the structure of the regression function hold. One key feature of the neural networks used in the…

We propose the BinaryGAN, a novel generative adversarial network (GAN) that uses binary neurons at the output layer of the generator. We employ the sigmoid-adjusted straight-through estimators to estimate the gradients for the binary neurons and train the whole network by end-to-end backpropogation. The proposed model …

Error backpropagation is a highly effective mechanism for learning high-quality hierarchical features in deep networks. Updating the features or weights in one layer, however, requires waiting for the propagation of error signals from higher layers. Learning using delayed and non-local errors makes it hard to reconcile…

In this work, we investigate the use of three information-theoretic quantities -- entropy, mutual information with the class variable, and a class selectivity measure based on Kullback-Leibler divergence -- to understand and study the behavior of already trained fully-connected feed-forward neural networks. We analyze …

We propose doubly nested network(DNNet) where all neurons represent their own sub-models that solve the same task. Every sub-model is nested both layer-wise and channel-wise. While nesting sub-models layer-wise is straight-forward with deep-supervision as proposed in \cite{xie2015holistically}, channel-wise nesting has…