Survey on reducing deep neural network training complexity.
problem Efficient training of large deep neural networks.
method Pruning and freezing parts of the network during training.
result Dimensionality reduction improves training efficiency.
For neural networks (NNs) with rectified linear unit (ReLU) or binary activation functions, we show that their training can be accomplished in a reduced parameter space. Specifically, the weights in each neuron can be trained on the unit sphere, as opposed to the entire space, and the threshold can be trained in a boun…
Local AdaAlter reduces communication in SGD with adaptive learning rates.
problem Communication overhead in distributed training.
method Novel SGD variant with adaptive learning rates and reduced communication.
result Empirically reduces communication overhead by up to 30%.
SWAT reduces training efficiency and speed for CNNs.
problem Efficient training of neural networks on sparse hardware.
method SWAT modifies backpropagation to eliminate small weights and activations.
result SWAT reduces FLOPS by 80% on ResNet-50 for ImageNet, improving training speed by 3.3x.
Meta-learning reduces training overhead for communication systems.
problem Inefficiency in machine learning due to frequent retraining when system configuration changes.
method Meta-learning selects a suitable inductive bias from related tasks, reducing training data and time requirements.
result Meta-learning can reduce training overhead for communication systems.
In this paper we deal with the offline handwriting text recognition (HTR) problem with reduced training datasets. Recent HTR solutions based on artificial neural networks exhibit remarkable solutions in referenced databases. These deep learning neural networks are composed of both convolutional (CNN) and long short-ter…
Deep learning reduces training overhead in massive MIMO systems.
problem Reducing training overhead in massive MIMO systems.
method Use of deep learning (NNs) to improve CSI acquisition and feedback processes.
result Significant improvements in performance and reduced complexity.
DRSSS method reduces model training costs by eliminating unsafe samples.
problem Reducing storage and training costs for customized models across different environments.
method Combines DR optimization and SSS for covariate shift.
result Models trained on reduced dataset perform similarly to those on full dataset.
Reduces risk of model inversion by reducing sensitive feature influence.
problem Model inversion attacks reveal sensitive individual data from trained models.
method Privacy-guided training to reduce sensitive feature influence in tree-based models.
result Training models to reduce sensitive feature influence reduces the risk of inference attacks.
New method reduces communication in deep learning training.
problem Communication overhead in distributed deep learning training.
method Random-block sparsification to reduce gradients communicated.
result Performance close to standard SGD with reduced communication.
The standard taxonomy of predictive uncertainty is inconsistent with standard measures.
problem Uncertainty taxonomy and measure inconsistency
method Proof of inconsistency
result Uncertainty is not reducible to data collection
Improved latent dynamics identification framework reduces training time and improves accuracy.
problem Accurate numerical solutions of partial differential equations require computationally expensive solvers.
method Sequential decoder training (mLaSDI) to correct residual errors from previous stages.
result mLaSDI consistently outperforms standard LaSDI, achieving lower prediction errors and reduced training time.
A new method reduces deep learning training costs by 92%.
problem High computational cost of backpropagation in deep neural networks.
method Dithered backprop with non-subtractive dither quantization.
result 92% sparsity with minimal accuracy loss.
We study the effect of the stochastic gradient noise on the training of generative adversarial networks (GANs) and show that it can prevent the convergence of standard game optimization methods, while the batch version converges. We address this issue with a novel stochastic variance-reduced extragradient (SVRE) optimi…
Real-time pruning during training reduces network size and training time.
problem Efficiently reducing neural network size without sacrificing accuracy.
method Activation density-based pruning during training.
result Up to 200x reduction in parameters and 60x reduction in inference compute operations.
New algorithms reduce communication in GNN training.
problem Higher communication costs in GNNs due to sparse connectivity.
method Parallel algorithms for sparse-dense matrix multiplication.
result Asymptotic reduction in communication compared to previous methods.
Memory is increasingly often the bottleneck when training neural network models. Despite this, techniques to lower the overall memory requirements of training have been less widely studied compared to the extensive literature on reducing the memory requirements of inference. In this paper we study a fundamental questio…
Paper applies FloatSD8 to LSTM networks, reducing complexity and power.
problem Training and inference complexity of LSTM networks.
method Applied FloatSD8 for weights, 8-bit quantization for gradients/activations, reduced arithmetic precision.
result Successfully trained LSTM models with reduced complexity and preserved accuracy.
PaGoDA reduces diffusion model training costs by 64x.
problem Diffusion models are computationally expensive during training.
method Three-stage pipeline: downsampled training, distillation, progressive super-resolution.
result PaGoDA achieves state-of-the-art performance with reduced training costs.
This work reduces computation cost for on-device CNN training.
problem High computation cost during on-device CNN training.
method Self-supervised instance filtering and error map pruning.
result Substantial computation saving without significant accuracy loss.
Structured pruning is a popular method for compressing a neural network: given a large trained network, one alternates between removing channel connections and fine-tuning; reducing the overall width of the network. However, the efficacy of structured pruning has largely evaded scrutiny. In this paper, we examine ResNe…
A new method reduces adversarial training time without overfitting.
problem Catastrophic overfitting in single-step adversarial training.
method FGSMPR: FGSM with PGD Regularization.
result Reduces the gap to multi-step adversarial training.
Existing methods for reducing the computational burden of neural networks at run-time, such as parameter pruning or dynamic computational path selection, focus solely on improving computational efficiency during inference. On the other hand, in this work, we propose a novel method which reduces the memory footprint and…
SpeakerStew verifies 46 languages with reduced training and inference costs.
problem Speaker verification for 46 languages with smart speaker interactions.
method Pooling multilingual data, triage between text-dependent and text-independent models.
result Training on multiple languages generalizes well and reduces computational requirements.
State-of-the-art convolutional neural networks (CNNs) used in vision applications have large models with numerous weights. Training these models is very compute- and memory-resource intensive. Much research has been done on pruning or compressing these models to reduce the cost of inference, but little work has address…
BayPOD-AL learns reduced-order models from high-fidelity data efficiently.
problem Capturing dynamics of complex systems with large training datasets.
method Bayesian active learning based on uncertainty-aware POD.
result BayPOD-AL reduces computational cost and improves model accuracy.
New method reduces model bias and variance by adjusting training sample weights based on label uncertainty.
problem Tradeoff between model bias and variance in classification models.
method Estimate label uncertainty, adjust training sample weights, and fine-tune decision boundary.
result Improves model performance and reduces variance in physical activity recognition.
Non-invasive myoelectric prostheses require a long training time to obtain satisfactory control dexterity. These training times could possibly be reduced by leveraging over training efforts by previous subjects. So-called domain adaptation algorithms formalize this strategy and have indeed been shown to significantly r…
Spatiotemporal sequence prediction is an important problem in deep learning. We study next-frame(s) video prediction using a deep-learning-based predictive coding framework that uses convolutional, long short-term memory (convLSTM) modules. We introduce a novel reduced-gate convolutional LSTM(rgcLSTM) architecture that…
We reduce training time in convolutional networks (CNNs) with a method that, for some of the mini-batches: a) scales down the resolution of input images via downsampling, and b) reduces the forward pass operations via pooling on the convolution filters. Training is performed in an interleaved fashion; some batches unde…
Paper optimizes neural network layers to reduce energy usage without sacrificing accuracy.
problem Energy consumption in deep neural networks during inference.
method Layerwise noise maximization to optimize reliability of memory elements.
result Reduces memory energy consumption by 3.3 times at equal accuracy.
PCRs compress data for deep learning, reducing training time.
problem Efficiently training deep learning models over large datasets.
method Combining progressive compression with an efficient storage layout.
result PCRs can tolerate up to 50% compression without significantly affecting training accuracy.
POET enables large neural network training on tiny devices with reduced energy.
problem Training large neural networks on memory-limited edge devices.
method Jointly optimizes rematerialization and paging for memory reduction, formulating an MILP for energy-efficient training.
result POET trains ResNet-18 and BERT within Cortex-M memory constraints, outperforming current methods in energy efficiency.
Designs a neural network to reduce training cost by mapping to higher dimensions.
problem High training cost in neural networks.
method Maps feature vectors to higher dimensional space, designs weight matrices to reduce cost, uses convex constraints.
result Reduces training cost as the number of layers increases, without cross-validation.
LCW reduces activation shift in neural networks, improving training efficiency and generalization.
problem Activation shift in neural networks leading to non-zero mean preactivation values.
method Linearly constrained weights (LCW) to reduce activation shift in fully connected and convolutional layers.
result LCW resolves the vanishing gradient problem and improves generalization of neural networks.
Recurrent neural networks can be difficult to train on long sequence data due to the well-known vanishing gradient problem. Some architectures incorporate methods to reduce RNN state updates, therefore allowing the network to preserve memory over long temporal intervals. To address these problems of convergence, this p…
A new RNN architecture reduces model size and improves performance.
problem Overparameterization and resource limitations in RNNs.
method Jointly encodes weight matrices using tensor-train factorization.
result Reduces model size by several orders of magnitude without sacrificing performance.
Distributed Lion optimizes large model training by reducing communication costs.
problem Training large AI models efficiently with reduced communication costs.
method Adapted Lion optimizer for distributed training, using binary or lower-precision vectors for communication.
result Distributed Lion achieves comparable performance to standard optimizers but with significantly reduced communication bandwidth.
Paper shows pre-training and transfer learning reduce sample complexity for neural networks.
problem Training high-dimensional supervised learning with limited labeled data.
method Study of single-layer neural networks via online stochastic gradient descent, considering concept shift.
result Pre-training and transfer learning reduce sample complexity by polynomial factors under general assumptions.
A new deep learning method using Boolean logic reduces training and inference energy.
problem High computational and energy costs in deep learning training and inference.
method Introduces Boolean weights and inputs for efficient training using Boolean logic.
result Achieves full-precision accuracy in ImageNet classification and surpasses state-of-the-art results in semantic segmentation.
This paper develops a method to train compact neural networks with reduced memory and computational costs.
problem Training large neural networks consumes excessive resources and energy.
method End-to-end training framework using Bayesian tensor decomposition with automatic rank determination.
result The method achieves significant parameter reduction and maintains or improves accuracy.
The Hausdorff Distance (HD) is widely used in evaluating medical image segmentation methods. However, existing segmentation methods do not attempt to reduce HD directly. In this paper, we present novel loss functions for training convolutional neural network (CNN)-based segmentation methods with the goal of reducing HD…
The well-known Mori-Zwanzig theory tells us that model reduction leads to memory effect. For a long time, modeling the memory effect accurately and efficiently has been an important but nearly impossible task in developing a good reduced model. In this work, we explore a natural analogy between recurrent neural network…
SIFT reduces training time by selecting samples with approximate losses.
problem Reducing training time by selecting samples with large approximate losses.
method Developed SIFT which uses early exiting to obtain approximate losses with intermediate layer representations for sample selection.
result SIFT achieves significant gains in training time and number of backpropagation steps without optimized implementation.
Neuron-specific dropout reduces overfitting and data needs for neural networks.
problem Overfitting and insufficient training data for deep neural networks.
method Compares training and validation passes of a layer, drops targeted neurons based on feature analysis.
result Achieves similar or better testing accuracy with less data, reducing overfitting.
Paper explores reducing precision in SVM for faster text classification.
problem Efficiency in multi-class text classification training.
method Comparison of SVM trained with reduced precision (16-bit, half) vs original.
result Reduced precision training maintains text classification accuracy.
Researchers enhance EfficientNet models for practical efficiency on Graphcore IPU.
problem Improving practical efficiency of EfficientNet models on high-performance accelerators.
method Group convolutions, proxy-normalized activations, and reduced training resolution.
result Improves practical efficiency for both training and inference on Graphcore IPU.
A scalable ROC-SVM variant reduces training time for imbalanced binary classification.
problem High computational cost of ROC-SVM for imbalanced binary classification.
method Incomplete U-statistics and low-rank kernel approximation.
result Comparable AUC performance with reduced training time.