Local SGD outperforms conventional methods in LLM training.
problem Training large language models on distributed devices.
method Local SGD algorithm applied to distributed training.
result Local SGD achieves competitive results in LLM training.
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
168,742 papers · 148 categories
Trend · papers per month
48 results for “local training”
Communication-Efficient Adaptive Batch Size Strategies for Distributed Local Gradient Methodsstat.ML
Adaptive batch sizes improve local gradient methods in distributed training.
problem Communication bottlenecks in distributed deep learning.
method Adaptive batch size strategies for local gradient methods.
result Adaptive batch sizes reduce minibatch gradient variance and improve training efficiency.
Today's state-of-the-art image classifiers fail to correctly classify carefully manipulated adversarial images. In this work, we develop a new, localized adversarial attack that generates adversarial examples by imperceptibly altering the backgrounds of normal images. We first use this attack to highlight the unnecessa…
Localized diffusion models reduce training complexity by exploiting low-dimensional structure.
problem Training diffusion models is computationally expensive due to the curse of dimensionality.
method Localized neural networks and localized score matching loss to estimate low-dimensional score functions.
result Localized diffusion models can circumvent the curse of dimensionality with reduced sample complexity.
The Generative Adversarial Networks (GANs) have demonstrated impressive performance for data synthesis, and are now used in a wide range of computer vision tasks. In spite of this success, they gained a reputation for being difficult to train, what results in a time-consuming and human-involved development process to u…
This paper proposes an alternative to E2E training for deep networks, reducing memory footprint.
problem High GPUs memory footprint in end-to-end training of deep networks.
method Locally supervised learning with information propagation loss to avoid information collapse.
result The proposed method achieves competitive performance with less than 40% memory footprint compared to E2E training.
Training-Time Batch Normalization Reshapes Local Partition Geometry in Piecewise-Affine Networkscs.LG
BN refines local partition geometry in piecewise-affine networks during training.
problem Understanding the effect of BN on the function realized during training in piecewise-affine networks.
method Analyzing the geometry of switching hyperplanes and affine-region partition conditioned on a mini-batch.
result BN increases expected local partition refinement in ReLU and piecewise-affine networks.
GradSkip: Communication-Accelerated Local Gradient Methods with Better Computational Complexitycs.LG
GradSkip reduces local training steps for better communication efficiency.
problem High communication costs in distributed optimization.
method GradSkip redesigns ProxSkip to allow clients with less important data to take fewer local training steps.
result GradSkip converges linearly with reduced local training steps and same accelerated communication complexity.
Study models deep learning training dynamics using locally elastic SDEs to reveal feature separability.
problem Understanding how deep learning models separate features from different classes during training.
method Modeling deep learning training using locally elastic SDEs with a drift term reflecting backpropagation impact.
result Local elasticity in SDEs leads to linear separability of features, resulting in vanishing training loss.
FALL improves local model training with anchor regularization.
problem Efficient local model training in regression tasks.
method Regularization with precomputed anchor models, closed-form solution.
result FALL outperforms network Lasso in accuracy with significantly less training time.
Local Coverage Governs Memorization in Diffusion Modelscond-mat.dis-nn
Local data coverage governs memorization in diffusion models.
problem Memorization in diffusion models
method Derive a theoretical criterion based on local data coverage
result Predicts memorization based on density of training data in neighborhood and dataset size
Proposes ContSup to boost local learning by supplying context between isolated modules.
problem Local learning's performance degrades with more isolated modules.
method Theoretical analysis and ContSup scheme to supply context between modules.
result Significant performance improvement with minimal overhead.
Backpropagation-free RL method trains layers using local signals.
problem Vanishing or exploding gradients in backpropagation-based RL.
method Local pairwise distance matching for layer-wise training without backpropagation.
result Backpropagation-free method achieves competitive performance and stability.
The problem of multi-speaker localization is formulated as a multi-class multi-label classification problem, which is solved using a convolutional neural network (CNN) based source localization method. Utilizing the common assumption of disjoint speaker activities, we propose a novel method to train the CNN using synth…
New techniques improve distributed training with compressed gradients.
problem Gradient mismatch problem in local error feedback.
method Step-ahead error feedback and error averaging techniques.
result Our methods handle gradient mismatch and train faster than full-precision training.
We present local discriminative Gaussian (LDG) dimensionality reduction, a supervised dimensionality reduction technique for classification. The LDG objective function is an approximation to the leave-one-out training error of a local quadratic discriminant analysis classifier, and thus acts locally to each training po…
Study local convergence of GDA for training GANs with kernel-based discriminators.
problem Analyzing the local dynamics of GDA for GANs with kernel-based discriminators.
method Linearization of a non-linear dynamical system, under an isolated points model assumption.
result Showed phase transitions indicating convergence, oscillation, or divergence of GDA.
No bad local minima: Data independent training error guarantees for multilayer neural networksstat.ML
We use smoothed analysis techniques to provide guarantees on the training loss of Multilayer Neural Networks (MNNs) at differentiable local minima. Specifically, we examine MNNs with piecewise linear activation functions, quadratic loss and a single output, under mild over-parametrization. We prove that for a MNN with …
The study proves sampling-based GNNs can approximate training on full graphs with small subgraphs.
problem Training Graph Neural Networks (GNNs) on large graphs is computationally expensive.
method Theoretical framework using graph local limits to prove approximation of GNN training on small samples.
result Parameters learned from sampling-based GNNs on small subgraphs are close to those on full graphs.
Local adaptive methods in FL can accelerate convergence but introduce bias, which is corrected.
problem The effect of using adaptive optimization methods for local updates in federated learning.
method Proposed correction techniques to overcome the bias introduced by local adaptive methods.
result Correction techniques can achieve faster convergence and higher test accuracy than baseline methods.
Federated learning improves with adaptive hyper-parameters and representation matching.
problem Heterogeneous client data leads to divergent local models in federated learning.
method Representation matching and adaptive hyper-parameters.
result Significant performance and robustness improvements in federated learning.
This paper proposes a novel approach to train deep neural networks by unlocking the layer-wise dependency of backpropagation training. The approach employs additional modules called local critic networks besides the main network model to be trained, which are used to obtain error gradients without complete feedforward …
We present ARU, an Adaptive Recurrent Unit for streaming adaptation of deep globally trained time-series forecasting models. The ARU combines the advantages of learning complex data transformations across multiple time series from deep global models, with per-series localization offered by closed-form linear models. Un…
Theory explains creativity in diffusion models generating novel images.
problem Diffusion models generate highly original images far from training data.
method Identified locality and equivariance as inductive biases to prevent optimal score-matching.
result Analytic models predict diffusion model outputs with high accuracy.
Local Gradient Descent with local steps converges to the centralized model in the interpolation regime.
problem Understanding the implicit bias of Local Gradient Descent in the interpolation regime.
method Analyzing the implicit bias of Local Gradient Descent for classification tasks with linearly separable data.
result The aggregated global model from Local-GD converges exactly to the centralized model in the interpolation regime.
New method certifies global robustness of neural networks efficiently.
problem Adversarial examples threaten certifiably robust neural networks.
method Formalized global robustness, adapted widely-used architectures with efficient global Lipschitz bounds.
result Certifiable robust models achieve state-of-the-art verifiable accuracy with negligible costs.
CutMix training technique improves spatial locality in Vision Transformers.
problem Improving spatial locality in Vision Transformers trained from scratch.
method Comparison of Baseline and Modern training protocols on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
result CutMix training component significantly reduces Mean Attention Distance (MAD) in early layers of Vision Transformers.
Federated learning algorithm reduces global model size by combining local and global representations.
problem Scalability issues in training large models on private data distributed over multiple devices.
method Proposes a federated learning algorithm that jointly learns compact local representations and a global model.
result The global model can be smaller since it only operates on local representations, reducing the number of communicated parameters.
Supervised training of neural networks for classification is typically performed with a global loss function. The loss function provides a gradient for the output layer, and this gradient is back-propagated to hidden layers to dictate an update direction for the weights. An alternative approach is to train the network …
Unified analysis of efficient local training methods for distributed variational inequalities.
problem Efficient distributed/federated learning for variational inequality problems.
method Unified convergence analysis of communication-efficient local training methods.
result First local gradient descent-accent algorithms with improved communication complexity.
A framework for federated adversarial learning with convergence analysis.
problem Unique vulnerabilities to adversarial attacks in federated learning.
method Formulates a general federated adversarial learning framework with inner and outer loops for client-side adversarial training and server-side model aggregation.
result The minimum loss under the proposed algorithm can converge to ε with chosen learning rate and communication rounds.
Federated learning performs distributed model training using local data hosted by agents. It shares only model parameter updates for iterative aggregation at the server. Although it is privacy-preserving by design, federated learning is vulnerable to noise corruption of local agents, as demonstrated in the previous stu…
Recent breakthroughs in computer vision make use of large deep neural networks, utilizing the substantial speedup offered by GPUs. For applications running on limited hardware, however, high precision real-time processing can still be a challenge. One approach to solving this problem is training networks with binary or…
FSL-Net detects and localizes feature shifts in large, high-dimensional datasets.
problem Feature shifts between data sources lead to erroneous features in various applications.
method FSL-Net is a neural network trained on multiple datasets to localize feature shifts.
result FSL-Net accurately localizes feature shifts from unseen datasets without re-training.
Study on local elasticity in neural network training, improving detection of class-specific changes.
problem Improving the detection of class-specific changes in neural network training.
method Comprehensive study of local elasticity, proposing a new definition to address limitations.
result New definition of local elasticity more sharply detects class-specific changes in neural network training.
We propose a novel model for temporal detection and localization which allows the training of deep neural networks using only counts of event occurrences as training labels. This powerful weakly-supervised framework alleviates the burden of the imprecise and time-consuming process of annotating event locations in tempo…
Discriminative Bimodal Networks for Visual Localization and Detection with Natural Language Queriescs.CV
Associating image regions with text queries has been recently explored as a new way to bridge visual and linguistic representations. A few pioneering approaches have been proposed based on recurrent neural language models trained generatively (e.g., generating captions), but achieving somewhat limited localization accu…
Delta-AI speeds up inference in sparse PGMs by local credit assignment.
problem Efficient inference in sparse probabilistic graphical models.
method Local credit assignment in agent's policy learning objective.
result Trained sampler recovers marginals and conditional distributions.
Personalized federated learning improves model accuracy and fairness by leveraging shared representations and local memorization.
problem Sub-optimal performance of federated learning when client data distributions are heterogeneous.
method Proposes a personalization mechanism based on local memorization of shared representations from a collectively trained global model.
result Significantly higher accuracy and fairness compared to state-of-the-art methods.
Efficient regularization mitigates catastrophic overfitting in single-step adversarial training.
problem Catastrophic overfitting in single-step adversarial training.
method ELLE regularization term to enforce local linearity of the loss function.
result Our regularization term effectively mitigates catastrophic overfitting without the drawbacks of previous methods.
Paper proposes a new strategy to improve initial performance of federated models.
problem Weight divergence in Federated Averaging (FedAvg) leads to poor initial performance in federated models.
method Local continual training with importance weights evaluated on a proxy dataset.
result The method significantly improves the initial performance of federated models with minimal extra communication costs.
Overlap-Local-SGD improves distributed SGD by overlapping communication and computation.
problem High communication delay and node slowdown in distributed SGD.
method Adding an anchor model to synchronize local updates and pull them towards the anchor model.
result Overlap-Local-SGD speeds up distributed training and mitigates straggler effects.
This paper evaluates and compares gradient leakage attacks in federated learning.
problem Gradient leakage attacks compromise client privacy in federated learning.
method Formal and experimental analysis of gradient leakage attacks, evaluation of attack effectiveness and cost.
result Gradient leakage attacks can reconstruct private local training data from shared parameter updates.
We design a non-convex second-order optimization algorithm that is guaranteed to return an approximate local minimum in time which scales linearly in the underlying dimension and the number of training examples. The time complexity of our algorithm to find an approximate local minimum is even faster than that of gradie…
Federated learning is a recently proposed paradigm that enables multiple clients to collaboratively train a joint model. It allows clients to train models locally, and leverages the parameter server to generate a global model by aggregating the locally submitted gradient updates at each round. Although the incentive mo…
RelatIF selects more intuitive training examples for explaining model predictions.
problem Influence functions identify outliers as explanatory examples, leading to poor explanations.
method RelatIF separates global and local influence, optimizing for local relative to global effects.
result Examples selected by RelatIF are more intuitive than those from influence functions.
We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from…
Local adaptation improves federated learning models.
problem Improving accuracy of federated learning models on non-iid data.
method Local adaptation techniques (fine-tuning, multi-task learning, knowledge distillation).
result Participants benefit from local adaptation, improving federated model accuracy.