Paper analyzes the free energy of CNNs with skip connections in Bayesian learning.
problem Dependency of CNNs with skip connections on the number of parameters.
method Examines the Bayesian free energy of CNNs with and without skip connections.
result The upper bound of free energy of Bayesian CNN with skip connections does not depend on overparametrization.
Skip connections make adversarial examples more transferable in ResNets.
problem Transferability of adversarial examples in ResNets with skip connections.
method Skip Gradient Method (SGM) to craft more transferable adversarial examples.
result SGM improves transferability of adversarial examples in various DNNs.
The paper proves skip connections help neural networks avoid shallow local minima.
problem Understanding how skip connections affect the loss landscape of deep neural networks.
method Theoretical analysis of the topology of loss landscapes of deep ReLU neural networks with skip connections.
result Skip connections help control the connectedness of sub-level sets, avoiding shallow local minima.
Skip connections improve biologically-inspired learning rules.
problem Biologically-inspired learning rules often underperform compared to backpropagation.
method Introduced skip connections between intermediate layers in biologically-motivated learning rules.
result Skip connections can match the performance of backpropagation and are robust to hyper-parameters.
Improved atomistic model predicts molecular properties using weighted skip-connections.
problem Understanding the relative importance of interactions in molecular property prediction.
method Extended SchNet architecture with weighted skip-connections to analyze molecule properties.
result Relative weighting of interaction blocks depends on molecule's chemical composition and configurational degrees of freedom.
New insights into GNN optimization reveal skip connections and depth accelerate training.
problem Understanding and optimizing the training of Graph Neural Networks (GNNs).
method Analysis of gradient dynamics in linearized GNNs and empirical validation.
result GNNs are implicitly accelerated by skip connections, more depth, and good label distribution during training.
This paper examines how skip connections prevent rank collapse in sequence models.
problem Rank collapse in sequence models, leading to reduced expressivity and training instabilities.
method Analytical and ablation studies of lambda-skip connections in SSMs.
result A sufficient condition to prevent rank collapse across various architectures.
This work analyzes how bottleneck layers and skip connections affect linear denoising autoencoders' generalization.
problem Understanding the generalization of linear denoising autoencoders in overparameterized regimes.
method Analyzes two-layer linear denoising autoencoders with a bottleneck layer and skip connection, deriving test risk formulas.
result Bottleneck layers introduce an additional complexity measure, while skip connections can mitigate variance.
DARTS- improves robustness by factoring out skip connections' advantage.
problem Performance collapse in DARTS architecture search.
method Factor out skip connections' advantage with an auxiliary skip connection.
result Significantly improved robustness across various datasets.
New metric reveals how topology affects deep network performance.
problem Understanding how topology influences deep network performance.
method Introduced NN-Mass metric to quantify gradient propagation and model performance.
result NN-Mass identifies models with similar accuracy but different sizes/compute requirements.
Theory explains why DARTS favors deep architectures over shallow ones.
problem DARTS selects architectures with dominated skip connections, leading to performance degradation.
method Theoretical analysis of operations' effects on network optimization; introduces sparse binary gates and path-depth-wise regularization.
result Theoretical proof that architectures with more skip connections converge faster.
Gradient descent finds global minima in deep neural networks with skip-connections.
problem Finding global minima in deep neural networks with skip-connections.
method Analysis of the gradient descent algorithm in over-parametrized deep neural networks with skip-connections.
result Gradient descent can find global minima exponentially fast in the over-parametrized regime.
Study on autoencoder denoising in high dimensions.
problem Denoising data from Gaussian mixtures.
method Two-layer non-linear autoencoder with skip connection in high-dimensional limit.
result Closed-form expressions for denoising mean-squared test error.
This paper explains GCNs using NTKs and improves their performance.
problem GCNs' performance degrades with depth, and skip connections marginally improve it.
method Derive NTKs for GCNs, validate with simulations, propose NTK as a surrogate model.
result Suitable normalisation can prevent drastic performance drop with depth.
Novel approach characterizes deep neural networks at initialization.
problem Characterizing the behavior of deep neural networks at initialization.
method A novel approach considering the evolution of statistical moments of signal and noise.
result Established that skip-connections in residual networks lead to well-behaved moments and no pathology.
A long-standing obstacle to progress in deep learning is the problem of vanishing and exploding gradients. Although, the problem has largely been overcome via carefully constructed initializations and batch normalization, architectures incorporating skip-connections such as highway and resnets perform much better than …
DVNet efficiently segments large neurovascular datasets using skip connections.
problem Challenges in segmenting large neurovascular datasets from high-throughput microscopy data.
method A fully-convolutional, deep, and densely-connected encoder-decoder network with skip connections.
result DVNet achieves superior performance in semantic segmentation of neurovascular datasets.
CAggNet improves medical image segmentation by fusing coarse and fine features.
problem Medical image segmentation accuracy and efficiency.
method Crossing Aggregation Network with nested skip connections and weighted aggregation.
result CAggNet achieves more accurate and efficient segmentation compared to existing methods.
SCARLET-NAS improves neural architecture search by stabilizing feature inconsistencies.
problem Difficulty in ranking subnetworks due to feature inconsistency in skip connections.
method Introduced an equivariant learnable stabilizer to homogenize feature disparities.
result SCARLET-A achieves 76.9% top-1 accuracy on ImageNet.
Proposes PSCs for UQ in deep nets without retraining.
problem Estimating uncertainty in deep nets with a single pass.
method Identifies sensitive, smooth intermediate layer, fits probabilistic model.
result PSCs achieve UQ and OOD detection performance matching existing methods.
Theoretical limits of deep residual networks show consistent covariance structures.
problem Understanding the limits of deep residual networks.
method Analyzing the behavior of deep residual networks with skip connections as width and depth approach infinity.
result Theoretical analysis confirms that the covariance structure remains consistent regardless of the order of width and depth.
Paper studies ResNet dynamics using NTH, reducing width requirement.
problem Understanding ResNet dynamics and improving training efficiency.
method Uses Neural Tangent Hierarchy (NTH) to analyze ResNet dynamics.
result Reduces width requirement from quartic to cubic for ResNet.
While enormous progress has been made to Variational Autoencoder (VAE) in recent years, similar to other deep networks, VAE with deep networks suffers from the problem of degeneration, which seriously weakens the correlation between the input and the corresponding latent codes, deviating from the goal of the representa…
Paper introduces a novel traffic forecasting model using autoencoders and exogenous variables.
problem Traffic forecasting using aggregated data from vehicles and infrastructure.
method Recurrent Autoencoder with skip connections and exogenous variables for dynamic traffic data.
result Model predicts speed, volume, and traffic direction with exogenous variables like weather and time.
OSA overcomes instability in skipless Transformers.
problem Instability in skipless Transformers using Softmax Self-Attention.
method OSA parametrizes attention matrix to be orthogonal via skew-symmetric matrix exponential.
result OSA allows for training non-causal Transformers without skip connections and normalisation layers.
Wide CNNs outperform infinite width networks, revealing scaling laws.
problem Understanding the performance difference between finite and infinite width convolutional networks.
method Diagrammatic approach to derive asymptotic width dependence for various quantities.
result The difference in performance between finite and infinite width models vanishes at a definite rate with respect to model width.
New framework explains deep learning using signal processing techniques.
problem Lack of a rigorous mathematical theory explaining deep learning performance.
method Transform-domain sparse regularization, Radon transform, and approximation theory.
result Explains neural network properties and performance.
Random feature maps improve forecasting of chaotic dynamical systems.
problem Forecasting chaotic dynamical systems with high accuracy.
method Data-driven random feature maps with tanh activation, skip connections, and localization.
result Effective forecasting skill for dynamical systems with dimensions up to 512.
Graph neural networks improve molecular property prediction.
problem Efficiently predicting molecular properties with high accuracy and scalability.
method Gated Graph Recursive Neural Networks (GGNN) with skip connections.
result GGNN achieves state-of-the-art performance on molecular property prediction benchmarks.
Deep neural networks can approximate functions with varying complexity.
problem Understanding the expressivity of deep neural networks in terms of function approximation.
method Using approximation theory, the study measures complexity by connections or neurons, and defines approximation spaces.
result Deep neural networks can approximate functions with low Besov smoothness if sufficiently deep, even with ReLU activation.
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…
NoisyDARTS injects random noise to improve neural architecture search.
problem Performance collapse in Differentiable Architecture Search (DARTS).
method Inject unbiased random noise to skip connections to impede gradient flow.
result NoisyDARTS achieves state-of-the-art results across various tasks.
Deep residual networks trained with gradient descent have small generalization gap.
problem Limited theoretical understanding of why residual networks generalize well.
method Analyzing overparameterized deep residual networks trained by gradient descent.
result Demonstrates that residual networks have a small generalization gap between training and test error.
New research shows non-bottlenecked autoencoders can outperform bottlenecked ones for anomaly detection.
problem The necessity of a bottleneck in autoencoders for anomaly detection.
method Investigated two ways to remove bottlenecks: overparameterising the latent layer and introducing skip connections. Carried out extensive experiments on various AE types and datasets.
result Non-bottlenecked autoencoders can outperform bottlenecked ones, improving anomaly detection performance.
Proposes Π-Nets, polynomial neural networks, for improved representation power.
problem Improving representation power in deep learning models.
method Introduces Π-Nets, a new class of deep polynomial neural networks. result Demonstrates Π-Nets outperform standard DCNNs and achieve state-of-the-art results. New framework explains deep neural networks using variational spline theory.
problem Understanding functions learned by deep neural networks.
method Developed a variational framework and function space.
result Deep ReLU networks are solutions to regularized data fitting problems over the proposed function space.
DNAS disentangles neural architecture search for better interpretability and performance.
problem Lack of interpretability in existing neural architecture search methods.
method DNAS disentangles the hidden representation of the controller into semantically meaningful concepts.
result DNAS achieves state-of-the-art performance and competitive architectures.
New neural architectures with multivariate nonlinearities are optimal in function space.
problem Optimality of neural architectures with multivariate nonlinearities.
method Construction of Banach spaces via k-plane transform and sparsity-promoting norm, proving representer theorem. result Neural architectures with multivariate nonlinearities are optimal in function space.
Deep Heaviside networks are limited but can be improved with connections or linear neurons.
problem Limited expressivity of deep Heaviside networks.
method Including skip connections or linear activation neurons improves expressivity.
result Lower and upper bounds for VC dimensions and approximation rates are derived.
Paper tackles dermoscopy lesion boundary detection using deep learning.
problem Automatic detection of lesion boundaries in dermoscopy images.
method Adapted U-net convolutional neural network with skip connections.
result Demonstrates effectiveness of deep learning in lesion boundary segmentation.
This paper explores how neural network width and depth behave as they approach infinity.
problem Understanding the behavior of neural functions as width and depth go to infinity.
method Formal definition of commutativity framework, study of neural covariance kernel, novel proof techniques.
result Taking width and depth to infinity in a deep neural network with skip connections results in the same covariance structure, regardless of the order of taking limits.
Deep generative models parameterized by neural networks have recently achieved state-of-the-art performance in unsupervised and semi-supervised learning. We extend deep generative models with auxiliary variables which improves the variational approximation. The auxiliary variables leave the generative model unchanged b…
We present a framework for translating unlabeled images from one domain into analog images in another domain. We employ a progressively growing skip-connected encoder-generator structure and train it with a GAN loss for realistic output, a cycle consistency loss for maintaining same-domain translation identity, and a s…
PS8-Net improves eight-state protein secondary structure prediction accuracy.
problem Precise prediction of eight-state protein secondary structure (PSS) is crucial in bioinformatics.
method PS8-Net is a new deep convolutional neural network (DCNN) that uses a PS8 module with skip connections to enhance accuracy.
result PS8-Net achieves 76.89% Q8 accuracy on benchmark datasets.
Deep ResNets can have better local minima than linear predictors.
problem Understanding the optimization landscape of deep ResNets compared to linear predictors.
method Analyzing the optimization landscape of ResNets with multiple residual blocks, showing geometric conditions under which ResNets have better local minima.
result Theorem showing that any critical point in the optimization landscape of deep ResNets is either at least as good as the best linear predictor or has a strictly negative eigenvalue in its Hessian.
NeuraLUT maps neural networks to lookup tables, reducing latency and improving expressivity.
problem Reducing latency in deep neural networks for FPGA accelerators.
method Mapping entire sub-networks to a single lookup table, introducing skip connections.
result Up to 4.3x lower latency for the same accuracy.
Empirical study finds deep learning assumptions often incorrect.
problem Widespread assumptions about neural networks are often incorrect.
method Empirical evaluation of neural network assumptions using various techniques.
result Proves existence of suboptimal local minima and demonstrates practical implications.
This work investigates how neural collapse improves transfer learning for large-scale models.
problem Improving transfer learning for large-scale models with limited labeled data.
method Investigates neural collapse and develops a fine-tuning method using skip-connections.
result Feature collapse on downstream data correlates with higher transfer accuracy.