NovoGrad improves deep learning training with adaptive moments and layer-wise normalization.
problem Training deep neural networks efficiently and effectively.
method Layer-wise adaptive moments with gradient normalization and decoupled weight decay.
result NovoGrad outperforms well-tuned SGD with momentum and Adam/AdamW in various tasks.
Study shows LLMs can remove half of layers without significant performance drop.
problem Understanding knowledge storage in LLMs' weights.
method Layer pruning and finetuning to identify and remove unnecessary parameters.
result Minimal degradation of performance after removing up to half of layers.
Layer rotation predicts model generalization, improving test accuracy by up to 30%.
problem Predicting model generalization in deep networks.
method Monitoring the cosine distance between layer weights and their initial values during training.
result Training procedures that maximize layer rotation consistently lead to better generalization performance.
This paper develops a multilayer spectral clustering method for heterogeneous data.
problem Clustering in multilayer graphs with varying layer weights and structures.
method Convex layer aggregation for multilayer spectral graph clustering (SGC).
result Phase transition analysis and automated cluster assignment with statistical guarantees.
Improved loss functions adapt to weight-space anisotropy, outperforming isotropic counterparts.
problem Adapting to the anisotropic nature of deep weight spaces for better performance.
method Refined local entropic loss functions restricted to a subset of weights, exploiting anisotropy.
result Partial local entropies outperform isotropic counterparts on image classification tasks.
Adapts neural network neurons' activation functions for better predictions.
problem Training neural networks with fixed activation functions limits their performance.
method Proposes training over a shape parameter, allowing neurons to adapt their own activation functions.
result Improves prediction accuracy by allowing neurons to tune their activation functions.
SPIRAL uses spikes to adaptively update weights, improving robustness and reducing overfitting.
problem Improving robustness and reducing overfitting in learning algorithms.
method Adaptive weight updates based on confidence estimates and activation offsets, regularized by spike rates.
result SPIRAL is more robust and less prone to overfitting compared to averaged perceptron and AROW.
Establishes connection between MTDNN and multitask GP, revealing weight correlation as key to task sharing.
problem Limited theoretical understanding of information sharing in MTDNN.
method Derives multitask GP kernels for MTDNN and MTBNN, showing shared hyper-parameters and last layer weights.
result Information sharing in MTDNN is due to weight correlation, not intermediate layer weights.
Global inducing points improve Bayesian neural network performance.
problem Improving Bayesian neural network performance.
method Adapting correlated approximate posterior to all layers in a Bayesian neural network and deep Gaussian processes using learned global inducing points.
result State-of-the-art performance on CIFAR-10 (86.7%) without data augmentation or tempering.
Self-configuring deep neural networks with fast training.
problem Training multi-level neural networks efficiently and automatically.
method Training neural networks layer by layer, using adaptive and self-adjusting parameters.
result Ability to self-configure and automatically build deep neural networks.
A new method improves few-shot image classification by updating top layers.
problem Few-shot image classification with limited data.
method Layer-wise adaptive updating (LWAU) for meta-learning.
result LWAU outperforms existing methods with a clear margin and learns more efficiently.
New spectral clustering method for multi-layer networks improves accuracy.
problem Detecting community structure in multi-layer networks.
method Integrative spectral clustering based on adaptive layer aggregation.
result Our methods minimize mis-clustering error and outperform existing methods.
We identify and approximate weights of two-layer neural networks from few samples.
problem Identifying and approximating weights of two-layer neural networks from limited data.
method Active sampling of finite difference approximations to Hessians, solving robust nonlinear programs, and gradient descent.
result Stable recovery of network weights under verifiable conditions.
Improved backpropagation with consequentialism weight updates for neural networks.
problem Improving backpropagation for neural networks, especially with mini-batch training.
method Introducing consequentialism weight updates derived from NLMS for multi-layer neural networks.
result The proposed method outperforms traditional BP and mini-batch training.
A new optimizer for deep learning improves accuracy and reduces training time.
problem Training deep neural networks for classification tasks.
method Hybrid Newton/Gradient Descent (NGD) method exploiting convexity of cross-entropy loss.
result Improves validation error and provides qualitative differences in hidden layer basis functions.
DFS dynamically decides bitwidths for layers to balance accuracy and efficiency.
problem Balancing model accuracy and inference speed for deep networks.
method Dynamic Fractional Skipping (DFS) framework that assigns bitwidths to layers for input-adaptive inference.
result DFS achieves superior tradeoff between computational cost and model accuracy.
LAWN normalizes logits to improve deep network adaptability and generalization.
problem Large logits and weights lead to overfitting in deep networks.
method Logit Attenuating Weight Normalization (LAWN) constrains weight norms in the final sub-network.
result LAWN improves generalization and adaptability of deep networks.
New neuron model learns and adapts its receptive field.
problem Learning and focusing on informative inputs.
method Adaptive locally connected neuron model using backpropagation.
result Focusing neurons outperform dense layers in classification tasks.
Adapting physics methods to data science for efficient feature learning.
problem Learning relevant features from large datasets efficiently.
method Layered tree tensor networks that scale linearly with data dimensions and training set size. Uses unsupervised learning for most layers and supervised learning for the top layer.
result Supervised classification of MNIST and fashion-MNIST datasets with good performance using fewer features.
WM layer improves CNN robustness to noise and adversarial attacks.
problem CNNs' susceptibility to noise and adversarial attacks.
method WM layer as a generic architectural addition to CNNs.
result WM variants enhance robustness to noise and adversarial attacks.
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.
Adaptively sparse Transformers improve interpretability and diversity in NLP.
problem Standard Transformers use dense attention, limiting interpretability and diversity.
method Introduces adaptively sparse Transformers using α-entmax for context-dependent sparsity. result Improves interpretability and diversity in NLP tasks without sacrificing accuracy.
Adaptive weight-decay improves deep neural network performance.
problem Overfitting in deep neural networks.
method AdaDecay adjusts weight-decay adaptively based on gradient norms within each layer.
result AdaDecay improves generalization and accuracy across various datasets and models.
AECF improves multimodal inference robustness and calibration.
problem Robustness and calibration issues in multimodal systems with missing inputs.
method Adaptive Entropy-Gated Contrastive Fusion (AECF) layer.
result Improves masked-input mAP by +18 pp at a 50% drop rate.
Adaptive neural networks cut inference time by 2.8x with minimal accuracy loss.
problem Efficiently evaluate deep neural networks for new examples without sacrificing accuracy.
method Two adaptive schemes: early exit and network selection, learned through binary classification.
result Dramatic reductions in computational cost with minimal accuracy loss.
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.
A new method compresses deep neural networks by predicting and quantizing weights between layers.
problem Resource constraints in deep neural networks.
method Inter-Layer Weight Prediction (ILWP) and quantization based on Smoothly Varying Weight Hypothesis (SVWH).
result The method achieves higher weight compression rates at the same accuracy level.
LoRAs enable efficient adaptation of large models; this paper explores processing LoRA weights with machine learning.
problem Efficient processing of low-rank weight decompositions in large finetuned models.
method Developed symmetry-aware invariant and equivariant LoL models to process LoRA weights.
result LoL models can predict CLIP scores, finetuning data attributes, and accuracy on downstream tasks.
New method connects neural networks to diagrammatic algebra.
problem Constructing permutation equivariant neural networks.
method Schur-Weyl duality between symmetric group and partition algebra.
result Simple diagrammatic method for calculating weight matrices.
Derives equations for deep learning biases and weights, showing data complexity reduction.
problem Understanding interpretability in supervised learning.
method Gradient flow equations and dynamical truncation of training data.
result Data complexity reduction at an exponential rate with training.
LoRA fine-tuning creates intruder dimensions that can cause forgetting, and a new law predicts when this happens.
problem Predicting when LoRA fine-tuning creates intruder dimensions that can cause catastrophic forgetting.
method Derived a per-layer critical update strength s∗ and an exact secular-equation characterization of the updated spectrum. result The law localizes the empirical threshold within a factor of two on 82% of layers and separates intruder-bearing from intruder-free layers at deployment.
End-to-end image super-resolution using Attention-based DenseNet with residual deconvolution.
problem Challenging task of improving low-resolution images.
method Proposes a novel ADRD model with weighted dense blocks and spatial attention modules.
result Demonstrates promising performance on publicly available datasets.
Weight normalization and reparametrized gradient descent adaptively regularize weights and converge to minimum l2 norm solutions.
problem Adapting to non-convex weight normalization for convergence to minimum l2 norm solutions.
method Weight normalization and reparametrized projected gradient descent (rPGD) for overparametrized least-squares regression.
result rPGD converges close to the minimum l2 norm solution, even for far-from-zero initializations.
Formula found for neural network error with fixed weights.
problem Understanding error in neural networks with fixed weights.
method Provided an explicit formula for approximation error.
result Explicit formula for neural network error with fixed weights.
BCGD algorithm improves training of quantized neural networks.
problem Training quantized deep neural networks at low bit-widths.
method Introduces coarse gradient descent and blended correction for training.
result BCGD achieves high accuracy in quantized neural networks.
N-BEATS-MOE improves time series forecasting by adapting to series characteristics.
problem Forecasting heterogeneous time series with varying characteristics.
method Mixture-of-Experts layer with dynamic block weighting.
result Consistent improvements across 12 benchmark datasets, especially for heterogeneous series.
Adapts BP-based algorithms for deep learning, improving performance and accuracy.
problem Training deep neural networks with discrete weights and activations.
method Message-passing algorithms based on Belief Propagation, with reinforcement field.
result Comparable performance to SGD-inspired heuristics (BinaryNet) and higher accuracy in predictions.
Adaptive learning of sample weights for better model performance.
problem Overfitting to biased training data with corrupted labels or class imbalance.
method Adaptive learning of an explicit weighting function using a meta-weight-net.
result Improves model accuracy in class imbalance and noisy label cases.
DSCF-Net learns deep features for clustering with robustness and locality preservation.
problem Unsupervised deep representation learning for clustering.
method Integrates robust deep concept factorization, deep self-expressive representation, and adaptive locality preserving feature learning.
result Delivers state-of-the-art performance on public databases.
A new method to improve deep neural networks using weight rescaling.
problem Overfitting and sensitivity to hyperparameters in weight decay.
method Weight rescaling (WRS) to control weight norm and prevent overfitting.
result WRS outperforms weight decay and other methods in various applications.
HyperLISTA simplifies LISTA training with adaptive hyperparameters.
problem Sparse recovery with LISTA networks.
method Adaptive hyperparameter tuning based on previous layers.
result HyperLISTA achieves similar performance on seen data and better on unseen data.
BiTAT improves neural network quantization for edge devices by focusing on weight dependencies and disentangling them.
problem Performance degradation of compact neural networks under extreme quantization.
method Task-dependent Aggregated Transformation (BiTAT) method that orthonormalizes weights and progressively quantizes them.
result BiTAT effectively preserves model performance on ImageNet and CIFAR-100 with compact backbones.
Max-pooling improves semantic segmentation by re-weighting under-represented classes.
problem Imbalanced training data distributions in semantic image segmentation datasets.
method Adaptive loss max-pooling that re-weights pixel contributions based on observed losses.
result Consistently improved semantic segmentation results on benchmark datasets.
Transfer learning improves ASR on limited resources.
problem End-to-end ASR training requires massive data and compute.
method Model adaptation using transfer learning.
result ASR models trained on limited resources achieve similar accuracy.
New bounds adaptively control spectral complexity of trained Transformers.
problem Understanding why Transformers generalize well in machine learning.
method Spectrum-adaptive post hoc generalization bounds for multi-layer Transformers.
result Bounds adaptively trade off spectral complexity against dimension and depth factors.
AdaRound improves post-training quantization of neural networks.
problem Improving the accuracy of quantized weights in neural networks.
method Adaptive rounding mechanism that adapts to data and task loss.
result AdaRound outperforms rounding-to-nearest and achieves state-of-the-art performance.
Transformers can emulate various algorithms by prompting, proving universality.
problem How to emulate algorithms using fixed-weight Transformers.
method Two modes of in-context algorithm emulation: task-specific and prompt-programmable. Constructing prompts that encode algorithm parameters into token representations.
result Fixed-weight Transformers can emulate a broad class of algorithms via prompts.
Optimizes neural networks' last layer with closed-form solutions.
problem Optimizing neural networks' last layer with stochastic gradient descent.
method Adapting closed-form last layer optimization for stochastic gradient descent, alternating between backbone and last layer updates.
result The method converges to optimal solutions and outperforms standard SGD and Adam in regression tasks.