Study infinite-depth limits of neural networks with fixed width.
problem Understanding the behavior of neural networks as depth increases with fixed width.
method Analyzing finite-width residual networks with random Gaussian weights, focusing on the infinite-depth limit.
result The pre-activations converge to a zero-drift diffusion process, differing from the infinite-width limit.
Study on feature learning dynamics in infinite-depth neural networks, focusing on ResNets.
problem Understanding how features evolve during training in deep neural networks, especially in the large-depth limit.
method Conditional Gaussian representations and SDE system with decoupled backward weights.
result Depth-induced suppression of forward-backward coupling in infinite-depth networks, leading to a decoupled forward-backward SDE system.
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.
Random neural networks have depth limits, which can be trained precisely.
problem Understanding the maximum depth of signal propagation in untrained neural networks.
method Mean field theory applied to randomly distributed weights and biases.
result Depth scales limit the maximum depth of trainable networks, and dropout destroys these limits.
A new depth measure and median defined on Hadamard manifolds.
problem Statistical depth and median on Hadamard manifolds.
method Horospherical depth and Busemann median defined using renormalized distance functions.
result The Busemann median exists for every Borel probability measure on Hadamard manifolds.
Study of deep linear neural networks with proportional width and depth.
problem Lack of descriptive power in Gaussian limit of deep linear neural networks.
method Proportional infinite-width infinite-depth limit for deep linear neural networks.
result Characterization of limiting distribution as a nontrivial mixture of Gaussians.
Study shows critical initialisation not crucial for ReLU networks under dropout limits.
problem Effect of initialisation on training speed and generalisation in ReLU networks.
method Large-scale statistical analysis of over 12,000 trained networks.
result Non-critical initialisations perform similarly to critical initialisations in terms of performance.
In order-driven markets, limit-order book (LOB) resiliency is an important microscopic indicator of market quality when the order book is hit by a liquidity shock and plays an essential role in the design of optimal submission strategies of large orders. However, the evolutionary behavior of LOB resilience around liqui…
ResNets approximate log-Gaussian at initialization, improving network performance.
problem Understanding the initialization behavior of deep neural networks like ResNets.
method Analyzing ReLU ResNets in the infinite-depth-and-width limit, showing log-Gaussian behavior.
result ResNets at initialization exhibit hypoactivation and interlayer correlations, which are not captured by Gaussian limits.
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.
One-shot neural architecture search limits depth search space and prunes networks for better performance and uncertainty.
problem Finding optimal depth in residual networks for efficient training and inference.
method Formulated a variational objective to approximate the depth distribution and pruned networks based on this distribution.
result Pruned networks achieve competitive accuracy with unpruned networks and better uncertainty calibration.
Study how depth affects inference in deep Bayesian neural networks.
problem Understanding how depth impacts inference in overparameterized linear Bayesian neural networks.
method Interpreting finite deep linear Bayesian neural networks as scale mixtures of Gaussian process predictors.
result Advances analytical understanding of how depth affects inference in a simple class of Bayesian neural networks.
Introduces Polar Depth for analyzing multivariate heavy-tailed data extremes.
problem Analyzing the behavior of extremes from multivariate heavy-tailed distributions.
method Introduces Polar Depth, a novel statistical depth function expressed in polar coordinates.
result The polar depth of the largest observations converges to the polar depth of the limiting distribution as the threshold increases.
New CNN learns depth features from scratch, outperforming transfer learning.
problem Limited depth data for RGB-D scene recognition.
method Bottom-up approach combining local weakly supervised training and global fine-tuning, modified CNN architecture.
result State-of-the-art accuracy on NYU2 and SUN RGB-D datasets.
Proposes a stochastic model for limit order book dynamics.
problem Captures the dynamics of limit order books in financial markets.
method Develops a stochastic partial differential equation (SPDE) model with multiplicative noise.
result Shows efficient estimation and computation methods for the model.
New insights into how depth and width affect in-context learning in deep models.
problem Understanding how various resources impact in-context learning in deep models.
method Analyzed linear regression in a deep linear self-attention model, varying resources like depth, width, context length, and training steps.
result Increasing depth improves in-context learning even at infinite context length, contrary to previous findings.
New algorithm proves deep networks can learn better than shallow ones.
problem Understanding the power difference between shallow and deep neural networks.
method Identifying a class of Boolean functions and proving that logarithmic-depth networks can learn them efficiently using hierarchical reconstruction.
result First algorithmic separation between constant-depth and logarithmic-depth neural networks.
New Transformer architecture prevents rank degeneracy in deep attention models.
problem Rank degeneracy in deep attention models.
method Modified Softmax-based attention model with skip connections, centered at identity, and scaled logits.
result Existence of a stable SDE implies well-behaved covariance structure, preventing rank degeneracy.
Residual networks with depthwise hyperparameter scaling transfer optimal hyperparameters across width and depth.
problem The challenge of hyperparameter tuning in deep learning, especially for large models.
method Combining μ μ μ P parameterization with residual networks having a residual branch scale of 1 / e x t d e p t h 1/\sqrt{ ext{depth}} 1/ e x t d e pt h . result Optimal hyperparameters transfer across width and depth in residual networks trained with this parameterization.
Study on learnability of deep random networks, showing practical limitations with depth.
problem Learnability of deep random networks with sign activation.
method Theoretical and practical analysis of random deep networks with sign activation.
result Learnability of random deep networks drops exponentially with depth.
Study on functions computed by deep-layered machines finds same distribution in neural networks and Boolean circuits.
problem Understanding the space of functions computed by deep-layered machines.
method Investigation of Boolean functions on random-layered machines, including neural networks and Boolean circuits.
result The space of functions computed at large depth limit is characterized and the macroscopic entropy of Boolean functions is either monotonically increasing or decreasing with depth.
Study examines infinite limits of transformer dynamics, identifying key parameterizations.
problem Understanding the training dynamics of transformer models in the feature learning regime.
method Analysis of infinite scaling limits using dynamical mean field theory.
result Identified parameterizations that admit well-defined infinite width and depth limits.
The paper proves barriers to approximating functions with small weights and depth in neural networks.
problem Proving barriers to approximating functions with constant depth neural networks.
method Reduction to open problems and natural-proof barriers in circuit complexity, and a new approach to polynomially-bounded functions.
result There are fundamental barriers to proving results beyond depth 4 for constant-depth neural networks.
Study on neural network initialization with shaped infinite depth-and-width networks.
problem Understanding the distribution of random covariance matrices in shaped infinite-depth-and-width networks.
method Introduced the Neural Covariance SDE to model the distribution of the random covariance matrix.
result Identified the precise scaling of the activation function necessary for a non-trivial limit.
This paper analyzes convergence rates of neural networks in the deep learning regime.
problem Understanding convergence rates of neural networks in the deep learning regime.
method Analyzing the Neural Tangent Kernel (NTK) convergence rates in the large depth limit.
result Quantifies the impact of initialization and activation function on NTK convergence rates.
Deep and wide ReLU networks learn data-dependent features even in the lazy training regime.
problem Understanding the behavior of neural networks with finite depth and width.
method Analyzing the mean and variance of the neural tangent kernel (NTK) in a randomly initialized ReLU network.
result The NTK has a non-trivial evolution during training, with the mean of its first SGD update being exponential in the ratio of depth to width.
Graphs can't learn certain tasks due to depth vs width limitations.
problem Understanding limitations of graph neural networks in learning specific tasks.
method Analyzing expressive power of GNNmp under depth, width, and node attributes.
result GNNmp can lose significant power when depth and width are restricted.
ResNets and DenseNets converge to NTK with depth and width, offering advantages for kernel regression.
problem Understanding convergence of ResNets and DenseNets to Neural Tangent Kernel (NTK).
method Analysis of finite width and depth corrections for NTK of ResNets and DenseNets.
result ResNets and DenseNets can converge to NTK with depth and width, unlike vanilla networks.
Stable ResNet stabilizes gradients in deep networks.
problem Gradient vanishing and exploding in deep ResNet architectures.
method Introducing Stable ResNet architectures with gradient stabilization and infinite depth expressivity.
result Stable ResNet maintains gradient stability and expressivity in deep networks.
Minimal width d i n + 1 d_{in}+1 d in + 1 allows ReLU nets to approximate any continuous function of d i n d_{in} d in variables.
problem Approximating continuous functions using ReLU nets with minimal width.
method Analyzing the expressive power of depth in neural nets with ReLU activations.
result Minimal width d i n + 1 d_{in}+1 d in + 1 is necessary and sufficient for ReLU nets to approximate any continuous function of d i n d_{in} d in variables. Uniform scaling limits in AdamW-trained transformers converge to ODEs.
problem Understanding the dynamics of large-depth transformers trained with AdamW.
method Modeling transformer dynamics as an interacting particle system coupled through attention, proving convergence to ODEs.
result The joint dynamics of hidden states and backpropagated variables converge uniformly to an ODE system.
Deep Gaussian processes can have non-degenerate and non-Gaussian limits.
problem Understanding the behavior of deep Gaussian processes as depth grows.
method Studying the limit of compositional Gaussian processes where each layer is a Gaussian process.
result Identified a sharp bandwidth threshold above which the limit is degenerate, and proved that for bandwidths below this threshold, the limit is a non-degenerate and non-Gaussian distribution.
Study on Tukey depth in machine learning using Hamilton-Jacobi equations.
problem Understanding Tukey depth in machine learning applications.
method Derive necessary conditions for Tukey depth in continuum limit, formulating them as a Hamilton-Jacobi equation.
result Prove existence and uniqueness of viscosity solutions for the derived equation, which bounds Tukey depth.
Dropout and RaM become equivalent in large ResNets as depth and width increase.
problem Improving performance in deep learning models.
method Comparing Dropout and Random Gradient Masking in ResNets.
result Dropout and RaM converge to the same large-scale limiting dynamics in ResNets.
Bayesian linear networks reveal optimal depth and width trade-offs.
problem Understanding how depth, width, and dataset size affect model quality in linear networks.
method Zero noise Bayesian inference with Gaussian weight priors and mean squared error.
result Optimal predictions at infinite depth and maximized Bayesian model evidence at infinite depth.
Improved self-supervised monocular depth estimation without complex architectures.
problem Challenges in acquiring per-pixel ground-truth depth data at scale.
method Proposed a set of improvements including a minimum reprojection loss, multi-scale sampling, and auto-masking loss.
result Surprisingly simple model leads to superior predictions compared to competing methods.
UDN adapts depth to data complexity, outperforming standard neural networks.
problem Adapting neural network depth to data complexity.
method Variational inference for infinitely deep neural networks with a novel algorithm.
result UDN outperforms standard neural networks and other infinite-depth approaches.
Study of infinitely deep but narrow neural networks using NTK theory.
problem Analyzing the role of depth in deep learning with overparameterized networks.
method Infinite-depth limit analysis of MLP and CNN using Neural Tangent Kernel (NTK) theory.
result Established trainability guarantee for infinitely deep but narrow neural networks.
Wide neural networks learn features under μ μ μ P, identifying weights and decomposing support.
problem Feature learning in wide neural networks under μ μ μ P. method Proving mean-field limit, characterizing identifiability, sparse-dictionary decomposition, and feature-learning-error decomposition.
result The triple ( w ∗ , D o r b ∗ , S ∗ ) (w^*, D^*_{\mathrm{orb}}, S^*) ( w ∗ , D orb ∗ , S ∗ ) identifies the natural learning cell of the architecture-data pair ( σ , ρ ) (σ, ρ) ( σ , ρ ) . Smooth activations enable optimal error rates in neural networks for Sobolev function classes.
problem Achieving optimal approximation and estimation error rates for neural networks in Sobolev function classes.
method Study of neural networks with smooth activations, proving optimal rates via approximation and statistical properties.
result Constant-depth networks with smooth activations achieve optimal rates of approximation and estimation, demonstrating smoothness adaptivity.
We study the analytical properties of a one-side order book model in which the flows of limit and market orders are Poisson processes and the distribution of lifetimes of cancelled orders is exponential. Although simplistic, the model provides an analytical tractability that should not be overlooked. Using basic result…
Single-pass method estimates neural network uncertainty.
problem Uncertainty estimation in deep learning requires multiple passes.
method Probabilistic reasoning over neural network depths.
result Single forward pass for uncertainty estimation.
The paper examines how the angle between inputs in ReLU networks decreases with depth, impacting training.
problem Depth degeneracy in neural networks, leading to constant function behavior on initialization.
method Combinatorial expansions and Monte Carlo experiments to analyze the angle between inputs in ReLU networks of increasing depth.
result The angle between inputs in ReLU networks decreases exponentially with depth, leading to constant function behavior on initialization.
Study on deep multi-head self-attention dynamics, proving homogenized limits under specific scalings.
problem Understanding the behavior of deep multi-head self-attention models as depth increases.
method Random model of deep multi-head self-attention, viewing depth as time, and analyzing the residual stream as a particle system.
result Homogenized limit of the dynamics, leading to deterministic or stochastic behavior depending on scaling, with implications for representation collapse.
Wide neural networks converge to Gaussian processes, improving generalization.
problem Understanding the generalization of wide neural networks, especially deep equilibrium models.
method Investigation of deep equilibrium models (DEQs) with infinite-depth layers, focusing on their convergence to Gaussian processes as width and depth approach infinity.
result Wide DEQs converge to Gaussian processes, maintaining generalization performance.
Wide neural networks with narrow bottlenecks behave like deep Gaussian processes.
problem Understanding the behavior of neural networks with narrow layers in the wide limit.
method Analyzing the wide limit of BNNs with narrow bottlenecks, showing they behave like a composition of GPs.
result Wide neural networks with narrow bottlenecks form a composition of GPs, termed a bottleneck NNGP.
Study reveals how Fisher information changes with network depth, finding it grows linearly.
problem Understanding the trainability of deep neural networks (DNNs).
method Investigates the spectral distribution of the conditional Fisher information matrix (FIM) for fully-connected networks achieving dynamical isometry.
result The conditional FIM's spectrum concentrates around the maximum and grows linearly with depth.
AR-GANs learn depth and DoF from unlabeled images using aperture rendering and focus cues.
problem Learning depth and DoF from unlabeled natural images with diverse viewpoints and shapes.
method Aperture rendering and focus cues to learn depth and DoF from unlabeled images.
result AR-GANs effectively learn depth and DoF from various datasets, including flower, bird, and face images.