SGD and weight decay encourage neural networks to learn low-rank weight matrices.
problem The bias of SGD towards low-rank weight matrices in neural networks.
method The study investigates the effect of SGD and weight decay on the rank of weight matrices in neural networks, both theoretically and empirically.
result Training with SGD and weight decay induces a bias towards rank minimization in weight matrices, which becomes more pronounced with smaller batch sizes and stronger weight decay.
Weight Decay induces low-rank weight matrices in neural networks, improving generalization.
problem Improving generalization in neural networks.
method Training ReLU NN with Weight Decay and Stochastic Gradient Descent.
result The weight matrix of a trained NN is approximately rank-two.
The paper analyzes how re-weighting helps in reducing variance in high-dimensional kernel methods under covariate shifts.
problem The challenge of high-dimensional kernel methods under covariate shifts and the role of re-weighting.
method Derives asymptotic expansion of high-dimensional kernels under covariate shifts, analyzes bias-variance decomposition, and characterizes the regularized kernel.
result Re-weighting helps in decreasing variance and can be seen as a data-dependent regularization.
Cross-validation under sample selection bias can, in principle, be done by importance-weighting the empirical risk. However, the importance-weighted risk estimator produces sub-optimal hyperparameter estimates in problem settings where large weights arise with high probability. We study its sampling variance as a funct…
Proposes stabilized weights for causal inference using isotonic calibration.
problem Stability and bias issues in inverse propensity weighting.
method Post-hoc isotonic calibration of inverse propensity weights.
result Improves performance of doubly robust estimators of average treatment effect.
Proposes volumization for neural networks to control bias-variance tradeoff.
problem Improving generalization and preventing memorization in neural networks.
method Defines a physical volume for weights, interpolating between L2 and L∞ regularization.
result Volumization interpolates between weight decay and clipping, improving generalization.
New CV method reduces bias in spatial prediction models.
problem Bias in standard cross-validation due to uneven sampling.
method Target-Weighted Cross-Validation (TWCV) framework.
result Weighted CV approaches reduce bias in prediction error.
A new method corrects weight values to improve treatment effect estimation.
problem Estimating heterogeneous treatment effects in high-dimensional data with sample selection bias.
method Differentiable Pareto-Smoothed Weighting (DPSW) framework.
result Our method outperforms existing methods in treatment effect estimation.
The paper connects neural collapse and low-rank bias in networks with L2 regularization.
problem Understanding the emergence of low-rank bias and neural collapse in L2-regularized networks.
method Unified theoretical framework linking TCV and rank of weight matrices, proving global optimality of DNC1, and establishing a benign landscape property.
result Zero TCV across intermediate layers minimizes representation cost under natural architectural constraints, and DNC1 is globally optimal.
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.
In importance sampling (IS)-based reinforcement learning algorithms such as Proximal Policy Optimization (PPO), IS weights are typically clipped to avoid large variance in learning. However, policy update from clipped statistics induces large bias in tasks with high action dimensions, and bias from clipping makes it di…
This work shows how penalising bias terms in norm regularisation leads to sparse solutions.
problem Understanding the relation between parameter norm regularization and the sparsity of neural network solutions.
method Analyzes one hidden ReLU layer networks with unidimensional data, showing the norm required for function representation and the importance of the bias term's norm.
result Penalising the bias terms in regularisation leads to sparse solutions, enforcing the uniqueness and sparsity of the minimal norm interpolator.
A learned generative model often produces biased statistics relative to the underlying data distribution. A standard technique to correct this bias is importance sampling, where samples from the model are weighted by the likelihood ratio under model and true distributions. When the likelihood ratio is unknown, it can b…
Improves transferability of representations from source to target domains with weights and invariant representations.
problem Label shift between source and target domains in unsupervised domain adaptation.
method Integrates weights and invariant representations to bound the target risk, highlighting the role of inductive bias.
result Empirical evidence shows that weak inductive bias makes adaptation more robust.
This work identifies and mitigates topological bias in HGNNs using meta-weighting and debiasing.
problem Topological bias in HGNNs affects model performance on specific nodes.
method Meta-weighting adjacency matrix, PageRank projection, debiasing structure.
result The debiasing structure improves HGNNs' performance and debiasing.
Optimizes weights for better model performance in shifting data.
problem Improper importance weighting leads to poor model performance in data shifts.
method Interprets weights as a bias-variance trade-off and optimizes them simultaneously with model parameters.
result Optimizing weights significantly improves model generalization performance.
New methods reduce bias in synthetic data for machine learning.
problem Statistical bias in synthetic data generated for privacy.
method Re-weighting strategies using privatised likelihood ratios.
result Private importance weighting enhances synthetic data utility.
Deep ResNets favor low bottleneck rank with proper hyperparameters.
problem Understanding the inductive bias of deep neural networks.
method Computed minimum-norm weights of a deep linear ResNet.
result Deep nonlinear ResNets have an inductive bias towards minimizing bottleneck rank.
Importance-weighting is a popular and well-researched technique for dealing with sample selection bias and covariate shift. It has desirable characteristics such as unbiasedness, consistency and low computational complexity. However, weighting can have a detrimental effect on an estimator as well. In this work, we empi…
This paper balances bias and variance in adaptive importance sampling using mirror descent.
problem Large variance in adaptive importance sampling weights.
method Regularization strategy with power raised importance weights connected to mirror descent.
result The regularization parameter balances bias and variance.
BR-SNIS reduces bias in self-normalized IS without increasing variance.
problem Bias in self-normalized IS.
method Iterated sampling-importance resampling (ISIR) to form a bias-reduced estimator.
result Significant reduction in bias without increasing variance.
Characterizes inductive bias in multi-channel linear CNNs with bounded weight norm.
problem Understanding the inductive bias in multi-channel linear convolutional networks.
method Function space characterization and empirical testing of gradient descent.
result The inductive bias depends on the number of output channels for multi-channel inputs but not for single-channel inputs.
Deep Reinforcement Learning improves with Weighted Q-Learning to reduce bias and uncertainty.
problem Overestimation and high variance in Q-Learning cause learning algorithms to diverge in complex environments.
method Deep Weighted Q-Learning (Deep WQL) uses Dropout and Monte Carlo sampling to approximate WQL's weights and reduce bias.
result Deep WQL reduces bias and improves performance on benchmarks compared to existing methods.
A new algorithm removes unexpected correlations in biased data for better clustering.
problem Clustering with selection bias in data.
method Decorrelation regularized K-Means (DCKM) algorithm.
result DCKM achieves significant performance gains on real-world datasets.
Recommendation systems have been integrated into the majority of large online systems to filter and rank information according to user profiles. It thus influences the way users interact with the system and, as a consequence, bias the evaluation of the performance of a recommendation algorithm computed using historical…
Paper proposes CounterSample to improve convergence in LTR models.
problem Large variance in IPS weights slows convergence in LTR models.
method Introduces CounterSample algorithm with provably better convergence.
result CounterSample converges faster than standard IPS-weighted methods.
Active learning introduces bias; this paper fixes it.
problem Bias in active learning due to non-representative training data.
method Formalized bias, identified situations where it's harmful/helpful, introduced corrective weights.
result Corrective weights can improve active learning, especially with overparameterized models.
Simplified analysis of SGD for linear regression with weight averaging.
problem Understanding SGD optimization in linear regression models.
method Simplified analysis using linear algebra tools, bypassing complex operator manipulations.
result Recovery of bias and variance bounds for SGD in linear regression.
This work explores the importance of model weights and Hessian bias in pruning.
problem Understanding the relative importance of model weights for efficient pruning.
method A principled exploration of pruning, focusing on linear models and neural networks.
result Asymptotic formulas reveal the performance of different pruning methods.
Mitigates bias in text classification by weighting instances.
problem Unintended biases in text classification datasets based on demographic terms.
method Instance weighting to recover non-discrimination distribution.
result Effective mitigation of unintended biases without sacrificing generalization.
Recommendation systems have been integrated into the majority of large online systems to filter and rank information according to user profiles. It thus influences the way users interact with the system and, as a consequence, bias the evaluation of the performance of a recommendation algorithm computed using historical…
Linear RNNs exhibit a bias towards shorter memory due to initialization variance.
problem Understanding the performance limitations of RNNs, especially linear ones.
method Kernel regime analysis to show equivalence to 1D-convolutional networks and analyze weightings.
result Linear RNNs with random initialization have a bias towards shorter memory periods.
We show how to convert ICL in linearized transformers into model weights.
problem Making in-context learning interpretable and permanent in large language models.
method Demonstrates equivalence between ICL and bias terms in linearized transformers, and develops ICLCA for exact conversion.
result Exact conversion of in-context learning into model weights is possible for linearized transformers.
Corrects sample selection bias in empirical risk minimization using importance sampling.
problem Statistical learning with biased training data.
method Weighted empirical risk minimization using importance sampling.
result Generalization capacity preserved with estimated importance weights.
Automates bias control in reinforcement learning algorithms.
problem Overestimation bias in reinforcement learning algorithms.
method Data-driven approach for automatic selection of bias control hyperparameters.
result Significant reduction in the number of interactions while maintaining performance.
Gradient descent on normalized networks reveals sparsity preferences.
problem Understanding the inductive bias of gradient descent on normalized neural nets.
method Analysis of gradient descent on weight-normalized smooth homogeneous neural nets, focusing on SWN and EWN.
result EWN causes weights to be updated in a way that prefers asymptotic relative sparsity.
Deep ReLU networks escape from the origin via saddle points with a low-rank bias.
problem Understanding the dynamics of gradient descent in deep ReLU networks.
method Analysis of escape directions and singular values of weight matrices.
result The first singular value of the ℓ-th layer weight matrix is at least ℓ41 larger than any other singular value. Paper proposes a new method combining random forests and Lasso selection.
problem Improving random forest performance by applying Lasso regression.
method Adaptive Lasso weighting applied to random forest predictions.
result Unified framework strictly outperforms other methods in simulations and real-world datasets.
AWNN improves matrix completion by adaptively weighting nearest neighbors.
problem Matrix completion with optimal nearest neighbor weights and radii selection.
method Adaptively weighted nearest neighbor method for matrix completion.
result Theoretical guarantees and synthetic experiments support the effectiveness of AWNN.
Algorithmic approaches endow deep learning systems with implicit bias that helps them generalize even in over-parametrized settings. In this paper, we focus on understanding such a bias induced in learning through dropout, a popular technique to avoid overfitting in deep learning. For single hidden-layer linear neural …
Study shows neural networks learn low frequencies first, proposing solutions.
problem Frequency bias in neural network learning process.
method Developed a PDE to unravel frequency dynamics, used Fourier Features model.
result Appropriate weight initialization can eliminate or control frequency bias.
Paper tackles exposure bias in recommender systems using contrastive learning.
problem Exposure bias in large-scale recommender systems.
method Contrastive learning to reduce exposure bias via inverse propensity weighting.
result Contrastive learning effectively reduces exposure bias in recommender systems.
TSC improves causal effect estimation in panel data.
problem Estimating causal effects in panel data with a single treated unit.
method Targeted synthetic control method that refines initial weights through a one-dimensional targeted update.
result TSC consistently improves estimation accuracy over state-of-the-art SCM baselines.
k-nearest neighbour (k-NN) is one of the simplest and most widely-used methods for supervised classification, that predicts a query's label by taking weighted ratio of observed labels of k objects nearest to the query. The weights and the parameter k∈N regulate its bias-variance trade-off, and the …
New measure corrects news bias in NLP stock return forecasting.
problem Improving stock return and volatility forecasting accuracy.
method Hype-Adjusted Probability Measure, sentiment score equation.
result Significantly improved forecast accuracy for U.S. semiconductor tickers.
CNNs can develop blind spots due to uneven padding in feature maps.
problem Spatial bias in convolutional networks leads to blind spots in certain tasks.
method Identified and analyzed the role of padding in convolutional networks, proposing solutions to mitigate bias.
result Mitigating spatial bias improves model accuracy, especially in tasks like small object detection.
Interpolated-MLPs control inductive bias for better performance in low-compute tasks.
problem Low-compute performance gap between MLPs and CNNs.
method Introduced Interpolated MLP (I-MLP) approach to control inductive bias incrementally.
result Continuous logarithmic relationship between inductive bias and performance in low-compute tasks.
Proposes methods to correct bias and missing data in regression models.
problem Nonignorable selection bias and missing response in regression models.
method Imputation-based and importance weighted regression methods, including repeated regression and doubly robust combination.
result Repeated regression can effectively correct bias and outperforms weighted regression in extrapolation.