Lectures on deep learning properties in infinite and large-width networks.
problem Understanding deep neural networks in extreme width conditions.
method Analysis of random deep neural networks, connections to linear models, kernels, and Gaussian processes, perturbative and non-perturbative treatments.
result Properties and behaviors of deep neural networks in the infinite-width limit and large-width regime.
Large deviation principle for deep neural networks with ReLU activation.
problem Understanding the behavior of deep neural networks with ReLU activation.
method Proving a large deviation principle for networks with Gaussian weights and ReLU activation functions.
result Simplified expressions and power-series expansions for the ReLU case.
IVON optimizes large neural networks, matching or outperforming Adam.
problem The inefficacy of variational learning in large neural networks.
method Improved Variational Online Newton (IVON) optimizer.
result IVON consistently matches or outperforms Adam for large networks.
EPINET improves neural networks with less computation.
problem Efficiently estimating uncertainty in neural networks.
method Introduces EPINET architecture that augments any neural network with minimal additional training.
result EPINET outperforms large ensembles with significantly less computation.
This work challenges the Neural Tangent Kernel's role in overparameterized neural networks, especially with large width and depth.
problem The Neural Tangent Kernel's behavior in overparameterized neural networks with large width and depth is unclear.
method Experimental and theoretical analysis of ReLU networks with large width and depth.
result The aggregate norm of hidden neuron deviations does not vanish in infinitely-wide ReLU networks, indicating non-trivial behavior.
We establish large deviation principles for convolutional neural networks.
problem Understanding the behavior of convolutional neural networks in the infinite-channel limit.
method We establish large deviation principles for convolutional neural networks under Gaussian prior and posterior distributions.
result We provide a large deviation principle for the sequence of conditional covariance matrices and the posterior distribution.
Quadratic models explain neural network behavior during training.
problem Understanding neural network dynamics during training with large learning rates.
method Developed and tested Neural Quadratic Models.
result Neural Quadratic Models exhibit the 'catapult phase' similar to neural networks.
Renormalized pruning improves neural network accuracy.
problem Over-parameterized neural networks waste many parameters.
method Propose renormalizing sparse neural networks.
result Renormalized pruning converges to zero error.
Study shows how large neural networks avoid overfitting through decoupling of feature learning and complexity growth.
problem Understanding inductive bias and generalization in large neural networks.
method Dynamical mean field theory applied to large two-layer networks.
result Training dynamics of large networks exhibit a separation of timescales, decoupling feature learning and overfitting.
NeuroMatch efficiently matches subgraphs in large graphs using neural networks.
problem Determining the presence and location of a query graph in a large target graph.
method NeuroMatch decomposes graphs into subgraphs, embeds them using graph neural networks, and matches them directly in the embedding space.
result NeuroMatch is 100x faster and 18% more accurate than existing methods.
New mathematical framework proves the effectiveness of reducing neural network sizes.
problem Selecting optimal neural network sizes to avoid overfitting.
method Adaptive group Lasso applied to one-hidden-layer feedforward networks.
result Adaptive group Lasso is consistent and can accurately reconstruct network sizes.
Crowdsourced training of large neural networks with decentralized Mixture-of-Experts.
problem Expensive training of large neural networks limits research contributions.
method Learning@home: decentralized Mixture-of-Experts for large, poorly connected participants.
result Performance and reliability of Learning@home surpass conventional distributed training.
Neural networks can separate non-separable data using feature maps.
problem Non-separable data in neural networks.
method Characterization of feedforward neural networks and use of feature maps.
result ReLU neural networks can separate concentric data.
Large learning rates improve neural network generalization, study shows.
problem Understanding why large learning rates lead to better neural network generalization.
method Visual analysis of training and testing loss landscapes, introduction of a nonlinear model.
result Extended phase with large learning rates leads to near-optimal generalization.
New neural network method simplifies high-dimensional data.
problem Scalability issues in nonlinear sufficient dimension reduction.
method Stochastic neural network with adaptive gradient algorithm.
result Proposed method outperforms existing methods on large-scale data.
The paper studies deep neural networks with Gaussian weights and finds their asymptotic behavior.
problem Understanding the behavior of deep neural networks with large width.
method Function-space perspective, Gaussian process analysis, weak convergence in large-width limit.
result Deep neural networks with large width converge to a continuous Gaussian process.
Bayesian neural networks explore rare fluctuations for better feature learning.
problem Understanding rare but dominant fluctuations in Bayesian neural networks.
method Large-deviation theory and joint optimization over predictors and internal kernels.
result Posterior rate function optimization reveals data-dependent kernel selection.
Dense neural networks learn efficiently with large datasets and noise.
problem Training neural networks with large, noisy datasets.
method Statistical mechanics and Monte Carlo simulations.
result Dense neural networks can handle large amounts of patterns and recognize patterns at high signal-to-noise ratios.
NTK theory fails to predict practical behavior of large-width neural networks.
problem Theoretical limits of NTK do not match practical neural network architectures.
method Empirical investigation of NTK's applicability to large-width architectures.
result Practically relevant behavior of large-width architectures differs from NTK theory.
New model explains deep learning performance at large learning rates.
problem Understanding deep learning performance at different learning rates.
method Developed neural networks with solvable training dynamics.
result Large learning rates lead to convergence to flatter minima.
Graph neural networks improve network localization accuracy and efficiency.
problem Network localization in large-scale networks.
method Adopted graph neural networks for nonlinear regression.
result GNN outperforms state-of-the-art benchmarks in network localization.
Large multilayer neural networks trained with backpropagation have recently achieved state-of-the-art results in a wide range of problems. However, using backprop for neural net learning still has some disadvantages, e.g., having to tune a large number of hyperparameters to the data, lack of calibrated probabilistic pr…
The paper proves neural networks with ReLU and softmax can approximate any function.
problem Approximating functions and class labels in neural networks.
method Extended universal approximator theory to neural networks with ReLU and softmax.
result Neural networks with ReLU and softmax can approximate any function and class labels.
We compress large neural networks for quick adaptation to specific contexts.
problem How to quickly adapt a pretrained large neural network to specific contexts.
method Propose a Bayesian hypernetwork framework to compress the network and encourage sparsity.
result Generated compressed networks are significantly smaller than baseline methods.
Large SGD step sizes lead to sparse feature learning in neural networks.
problem Sparse feature learning in neural networks with large step sizes.
method Empirical observations and theoretical analysis of SGD dynamics.
result Large step sizes induce implicit regularization leading to sparse predictors.
GPUPoly verifies large neural networks robustly on GPUs.
problem Proving robustness of deep neural networks is crucial but challenging.
method Custom polyhedra algorithms on GPUs.
result GPUPoly can verify 1M neuron networks in 34.5 ms.
Single neural network predicts ImageNet model parameters for faster training.
problem Training diverse ImageNet models requires significant resources and time.
method Trained a neural network to predict ImageNet model parameters and used them for initialization.
result Models initialized with predicted parameters converge faster and achieve competitive performance.
A new method uses deep neural networks for estimating individual treatment effects.
problem Estimating individual treatment effects in large models.
method Extended fiducial inference with Double Neural Network (Double-NN) method.
result The Double-NN method outperforms CQR in individual treatment effect estimation.
We analyze multi-layer neural networks in the asymptotic regime of simultaneously (A) large network sizes and (B) large numbers of stochastic gradient descent training iterations. We rigorously establish the limiting behavior of the multi-layer neural network output. The limit procedure is valid for any number of hidde…
Review of neural network expressivity and architectures.
problem Understanding neural network expressivity across different architectures.
method Comprehensive overview of approximation results for various neural network types.
result Deep neural networks offer advantages over shallow ones for specific function classes.
Graph neural networks improve El Niño forecasts.
problem Improving seasonal forecasting accuracy for El Niño-Southern Oscillation.
method Designing a novel graph connectivity learning module to model large-scale spatial interactions with ENSO forecasting.
result Our model \graphino outperforms state-of-the-art models for forecasts up to six months ahead.
Artificial neural networks with millions of adjustable parameters and a similar number of training examples are a potential solution for difficult, large-scale pattern recognition problems in areas such as speech and face recognition, classification of large volumes of web data, and finance. The bottleneck is that neur…
Enhanced image denoising with MWRDCNN using residual dense blocks.
problem Image denoising with improved performance and robustness.
method Multi-wavelet residual dense convolutional neural network (MWRDCNN) with residual dense blocks (RDBs).
result Significantly improved performance in image denoising compared to existing techniques.
Regularization is essential when training large neural networks. As deep neural networks can be mathematically interpreted as universal function approximators, they are effective at memorizing sampling noise in the training data. This results in poor generalization to unseen data. Therefore, it is no surprise that a ne…
Sparse Meta Networks adapt deep neural networks incrementally for fast learning.
problem Training deep neural networks is slow and impractical for complex, changing environments.
method Sparse Meta Networks use a memory layer to learn online sequential adaptation, accumulating fast-weights incrementally.
result Sparse Meta Networks achieve strong performance in various sequential adaptation scenarios.
Law derived for neural networks with sparse connections.
problem Understanding the behavior of neural networks with sparse connections.
method Law of large numbers for empirical distribution of parameters derived.
result Law for neural networks with sparse connections derived.
Analyzes unsupervised neural networks using statistical mechanics and Monte Carlo simulations.
problem Understanding computational capabilities of unsupervised neural networks.
method Statistical mechanics approach and Monte Carlo simulations.
result Obtained a phase diagram summarizing network performance.
This paper proposes a convolutional neural network (CNN)-based method that learns traffic as images and predicts large-scale, network-wide traffic speed with a high accuracy. Spatiotemporal traffic dynamics are converted to images describing the time and space relations of traffic flow via a two-dimensional time-space …
New insights into Hessian structure of neural networks reveal two forces.
problem Understanding the Hessian structure of neural networks.
method Analyzing the static and dynamic forces, comparing limit distributions using random matrix theory.
result The Hessian structure arises from a combination of static and dynamic forces, with C being a primary driver. Gradient-free deep learning for large datasets.
problem Training deep neural networks on large-scale datasets is resource-intensive and requires specialized techniques.
method Recursive Local Representation Alignment (RLRA) for gradient-free training.
result RLRA achieves comparable performance to backprop while converging faster and being parallelizable.
Deep neural networks (DNNs) have demonstrated dominating performance in many fields; since AlexNet, networks used in practice are going wider and deeper. On the theoretical side, a long line of works has been focusing on training neural networks with one hidden layer. The theory of multi-layer networks remains largely …
Next-gen reservoir computers fail to predict complex processes, highlighting need for better architectures.
problem Predicting complex, non-Markovian processes with recurrent neural networks.
method Lower bound from Fano's inequality and analysis of large probabilistic state machines.
result Next-generation reservoir computers have an error probability at least 60% higher than optimal for highly non-Markovian processes.
Efficiently applies NTK to large-scale datasets using random features.
problem Computational limitations of kernel methods for large-scale datasets.
method Proposes a sketching-based algorithm combining random features of arc-cosine kernels to construct an efficient feature map of the NTK.
result Achieves comparable error bounds to exact kernel methods but with significantly reduced feature dimensionality.
We analyze single-layer neural networks with the Xavier initialization in the asymptotic regime of large numbers of hidden units and large numbers of stochastic gradient descent training steps. The evolution of the neural network during training can be viewed as a stochastic system and, using techniques from stochastic…
SIAN bridges simple models to neural networks by identifying necessary feature combinations.
problem The gap between simple models and powerful neural networks in performance.
method Feature interaction detection and sparse selection algorithm.
result Competitive performance across multiple tabular datasets with optimal tradeoff.
Distributed implementations of mini-batch stochastic gradient descent (SGD) suffer from communication overheads, attributed to the high frequency of gradient updates inherent in small-batch training. Training with large batches can reduce these overheads; however, large batches can affect the convergence properties and…
Study rare-event simulation for neural networks and random forests.
problem Safety evaluation and robustness quantification of machine learning models.
method Importance sampling scheme integrating large deviations and sequential mixed integer programming.
result Efficiency guarantees and numerical demonstrations for various neural network architectures.
Deep learning has been widely applied and brought breakthroughs in speech recognition, computer vision, and many other domains. The involved deep neural network architectures and computational issues have been well studied in machine learning. But there lacks a theoretical foundation for understanding the approximation…