A new method for unsupervised domain adaptation using Gaussian processes.
problem Reducing target domain error by aligning input and output distributions.
method Max-margin Gaussian process approach to achieve hypothesis consistency.
result Our method effectively minimizes maximum discrepancy and maximizes margins.
This study analyzes adversarial training on linearly separable data and finds that gradient updates can achieve large margins in polynomial iterations.
problem Ensuring robustness in machine learning models trained on linearly separable data.
method Analysis of adversarial training with gradient updates on linearly separable data.
result Gradient updates in adversarial training can achieve large margins in polynomial iterations, whereas non-smooth methods require exponentially many iterations.
How can we efficiently propagate uncertainty in a latent state representation with recurrent neural networks? This paper introduces stochastic recurrent neural networks which glue a deterministic recurrent neural network and a state space model together to form a stochastic and sequential neural generative model. The c…
Derives asymptotic generalization error for large-margin classifiers.
problem Understanding the generalization error of large-margin classifiers.
method Statistical physics replica method for deriving asymptotic expression.
result Establishes phase transition boundary for class separability.
Paper improves DP-ERM for binary linear classification with large-margin subsets.
problem Differentially private binary linear classification with large-margin subsets.
method Efficient (ε,δ)-DP algorithm with empirical zero-one risk bound. result Improved empirical zero-one risk bound for binary linear classification.
Existing Bayesian models, especially nonparametric Bayesian methods, rely on specially conceived priors to incorporate domain knowledge for discovering improved latent representations. While priors can affect posterior distributions through Bayes' rule, imposing posterior regularization is arguably more direct and in s…
Determinantal point processes (DPPs) offer a powerful approach to modeling diversity in many applications where the goal is to select a diverse subset. We study the problem of learning the parameters (the kernel matrix) of a DPP from labeled training data. We make two contributions. First, we show how to reparameterize…
Improves MKL for multi-class classification with better feature selection and representation.
problem Real-world multi-class classification problems with non-linear separations.
method Large-margin multiple kernel learning (LMMK) with sparsity term for discriminative feature selection.
result Competitive classification accuracy and sparse non-zero kernel weights.
Cross-entropy loss together with softmax is arguably one of the most common used supervision components in convolutional neural networks (CNNs). Despite its simplicity, popularity and excellent performance, the component does not explicitly encourage discriminative learning of features. In this paper, we propose a gene…
Posterior regularization enhances Bayesian hierarchical mixture clustering by improving node separation.
problem High nodal variance in BHMC trees, leading to weak separation between nodes at higher levels.
method Employing Posterior Regularization to impose max-margin constraints on nodes at every level.
result Improves cluster separation in BHMC models, enhancing overall model performance.
Improved U-Nets with various intermediate blocks enhance singing voice separation.
problem Improving singing voice separation accuracy using U-Net architectures.
method Implemented and compared U-Nets with different intermediate spectrogram transformation blocks.
result A specific block type achieves state-of-the-art SDR by 0.9 dB.
New decision-theoretic characterization separates belief and decision posteriors.
problem Understanding the conditions under which loss-based updating coincides with Bayesian updating.
method Decision-theoretic approach to distinguish belief and decision posteriors.
result Generalized Bayes coincides with ordinary Bayesian updating only if the loss is proportional to negative log-likelihood.
New study shows non-interactive privacy model requires exponentially more data.
problem Privacy constraints limit data analysis efficiency.
method Developed new technique to prove lower bounds on data samples needed.
result Exponential lower bound on samples needed for learning tasks.
Paper establishes comparison theorems for large-margin learning.
problem Data piling issue in high-dimension and low-sample size SVM.
method Large-margin unified machines (LUM) loss functions.
result New comparison theorems for all LUM loss functions.
Introduces gradient decay in Softmax for better generalization.
problem Improving generalization performance in neural networks.
method Gradient decay hyperparameter in Softmax for varying gradient rates based on probability.
result Gradient decay rate affects generalization performance and can be tuned for better optimization.
New private algorithms learn large-margin halfspaces efficiently.
problem Learning large-margin halfspaces with privacy constraints.
method Differentially private algorithms based on a new approach.
result Sample complexity depends only on the margin, not dimension.
The paper studies conditions for exact posterior modeling in Bayesian networks.
problem Exact modeling of posterior distributions in Bayesian networks.
method Derives conditions for a recognition network to model the true posterior distribution exactly.
result Perfectness of the recognition network is crucial for local conditions to hold.
Study high-dimensional Bayesian linear regression using variational inference.
problem High-dimensional Bayesian linear regression with product priors.
method Non-linear large deviations theory and variational inference.
result Unique optimizer in variational problem governs posterior distribution under separation condition.
The key issue of few-shot learning is learning to generalize. This paper proposes a large margin principle to improve the generalization capacity of metric based methods for few-shot learning. To realize it, we develop a unified framework to learn a more discriminative metric space by augmenting the classification loss…
New method accelerates large margin metric learning for nearest neighbor classification.
problem Efficiently learning metrics for nearest neighbor classification.
method Triplet mining and stratified sampling for large margin metric learning.
result Improved efficiency and scalability of optimization.
Framework for private, noise-tolerant, and efficient learning algorithms.
problem Private and efficient learning of large-margin halfspaces in noisy environments.
method Simple framework using differential privacy and noise tolerance conditions.
result Noise-tolerant and private PAC learners for large-margin halfspaces with sample complexity independent of dimension.
Gradient-guided nested sampling improves posterior inference efficiency.
problem Efficiently sampling from complex posterior distributions.
method Gradient-guided nested sampling combining differentiable programming, Hamiltonian slice sampling, clustering, mode separation, dynamic nested sampling, and parallelization.
result Significantly faster mode discovery and more accurate partition function estimates.
Paper proposes a new classifier for hyperbolic spaces using horospherical boundaries.
problem Optimization of large margin classifiers in hyperbolic spaces.
method Horospherical decision boundaries for geodesically convex optimization.
result Geodesically convex optimization leads to globally optimal solutions.
We obtain a tight distribution-specific characterization of the sample complexity of large-margin classification with L_2 regularization: We introduce the γ-adapted-dimension, which is a simple function of the spectrum of a distribution's covariance matrix, and show distribution-specific upper and lower bounds on the s…
How can we learn, transfer and extract handwriting styles using deep neural networks? This paper explores these questions using a deep conditioned autoencoder on the IRON-OFF handwriting data-set. We perform three experiments that systematically explore the quality of our style extraction procedure. First, We compare o…
We derive and analyze a new, efficient, pool-based active learning algorithm for halfspaces, called ALuMA. Most previous algorithms show exponential improvement in the label complexity assuming that the distribution over the instance space is close to uniform. This assumption rarely holds in practical applications. Ins…
Bayesian nonparametric models get better posterior estimates via SPDE methods.
problem Estimating posterior distributions in nonparametric Bayesian models.
method Extending diffusion methods to SPDEs on Hilbert spaces for posterior contraction and Laplace approximation.
result Derivation of posterior contraction rates and finite-sample Bernstein von Mises results.
Paper provides first theoretical guarantees for hyperbolic space learning.
problem Learning a classifier in hyperbolic space for hierarchical data.
method Efficient algorithm for large-margin hyperplane learning in hyperbolic space.
result The low embedding dimension in hyperbolic space leads to superior classifier learning guarantees.
Improved acoustic scene classification with factorized CNN.
problem Acoustic scene classification in varying environments.
method Large-margin factorized CNN with triplet loss.
result Improved performance and better generalization on unseen data.
Bayesian inference improves neural network predictions by separating aleatoric and epistemic uncertainties.
problem Improving prediction accuracy of neural networks by quantifying and separating uncertainties.
method Approximated posterior distributions using deep ensembles for various neural network architectures.
result Prediction accuracy depends on both aleatoric and epistemic uncertainties, not just marginalized uncertainty.
Improved VAEs by training a contrastive prior to match posterior.
problem Prior hole problem in VAEs, leading to poor image generation.
method Introduced a contrastive energy-based prior and trained it using noise contrastive estimation.
result Significant improvement in VAE generative performance on various datasets.
Differential privacy of Gaussian process posterior sampling
problem Privacy of posterior sample paths from Gaussian process
method Intrinsic randomness yields DP guarantees
result Intrinsic randomness yields DP guarantees
Sampling from posterior distributions using Markov chain Monte Carlo (MCMC) methods can require an exhaustive number of iterations, particularly when the posterior is multi-modal as the MCMC sampler can become trapped in a local mode for a large number of iterations. In this paper, we introduce the pseudo-extended MCMC…
New findings show the large margins theory is insufficient for explaining ensemble methods.
problem Explaining the performance of ensemble methods, especially boosting.
method Illustrated by counterexamples that show how to improve margin distribution without improving test set performance.
result The large margins theory is not sufficient to explain the performance of ensemble methods.
We obtain a tight distribution-specific characterization of the sample complexity of large-margin classification with L2 regularization: We introduce the margin-adapted dimension, which is a simple function of the second order statistics of the data distribution, and show distribution-specific upper and lower bounds on…
Gradient penalty improves GAN performance by inducing a large-margin classifier.
problem Improving GAN performance and addressing vanishing gradients.
method A unifying framework of expected margin maximization, showing gradient penalties induce large-margin classifiers.
result Gradient penalties reduce vanishing gradients and produce better generated outputs.
Corrects approximate Bayesian inference for better decision-making.
problem Sub-optimal decisions due to inaccurate posterior predictive distributions.
method Trains a separate model to correct decision-making under approximate posterior, combining Bayesian modeling with optimization.
result Empirically demonstrates improved predictive accuracy in various problems.
Paper proposes LMM-PQS for cross-domain few-shot learning.
problem Cross-domain few-shot learning problem.
method Generates pseudo query images and fine-tunes feature extraction modules with a large margin mechanism.
result LMM-PQS outperforms baseline models in cross-domain few-shot learning.
Bayesian approach uses generative models as priors for better source separation.
problem Artifacts in source separation for richly structured data.
method Bayesian approach with generative models as priors and noise-annealed Langevin dynamics.
result Achieves state-of-the-art performance for MNIST digit separation.
A simple pooling technique prevents posterior collapse in sequence VAEs.
problem Posterior collapse in sequence VAEs, causing models to ignore latent variables.
method Proposed a pooling technique to prevent posterior collapse.
result Significantly better data log-likelihood achieved compared to standard sequence VAEs.
Markov chain Monte Carlo (MCMC) algorithms have become powerful tools for Bayesian inference. However, they do not scale well to large-data problems. Divide-and-conquer strategies, which split the data into batches and, for each batch, run independent MCMC algorithms targeting the corresponding subposterior, can spread…
Detecting and recovering labels in binomial logistic mixtures is challenging due to an information gap.
problem Detecting and recovering labels in binomial logistic mixtures
method Propose two feasibility-aware inference procedures
result Avoid misleading component selections and improve label probability calibration
SwISS improves scalability of Bayesian inference for large datasets.
problem Scalability issues in Bayesian inference for large datasets.
method Divide-and-conquer approach with SwISS for recombining sub-posterior samples.
result SwISS accurately approximates the original posterior distribution.
New online learning algorithm combines PA and TER for binary classification.
problem Binary classification with non-separable data and data imbalance.
method Online Passive-Aggressive (PA) and Total-Error-Rate (TER) learning combined into PATER algorithm.
result PATER algorithms outperform existing online learning algorithms in efficiency and effectiveness.
We consider the problem of detecting out-of-distribution images in neural networks. We propose ODIN, a simple and effective method that does not require any change to a pre-trained neural network. Our method is based on the observation that using temperature scaling and adding small perturbations to the input can separ…
This paper introduces a probabilistic framework for k-shot image classification. The goal is to generalise from an initial large-scale classification task to a separate task comprising new classes and small numbers of examples. The new approach not only leverages the feature-based representation learned by a neural net…
We present a new method for the separation of superimposed, independent, auto-correlated components from noisy multi-channel measurement. The presented method simultaneously reconstructs and separates the components, taking all channels into account and thereby increases the effective signal-to-noise ratio considerably…
Efficient algorithms improve learning of large-margin halfspaces.
problem Learning large-margin halfspaces efficiently and reproducibly.
method Design of efficient, dimension-independent, polynomial-time algorithms; SGD-based approach; DP-to-Replicability reduction.
result Improved sample complexity compared to previous algorithms, with optimal sample complexity for one algorithm.