arXiv research

Theory for RLHF generalization under reward shift and clipped KL.

This paper shows RL with KL penalties is equivalent to Bayesian inference for fine-tuning LMs.

Local mass perspective on Bayesian inference

Paper studies regularized KKL divergence for distributions with disjoint supports.

The paper introduces a new divergence measure for variational autoencoders to improve reconstruction and generation.

This paper improves inverse problem solving with weakly convex regularisers and proves convergence.

GNIs induce a regulariser that penalizes high-frequency components in neural network activations.

A Python package solves source duplication in single channel LVMs using spectral regularisation.

We consider a general regularised interpolation problem for learning a parameter vector from data. The well known representer theorem says that under certain conditions on the regulariser there exists a solution in the linear span of the data points. This is the core of kernel methods in machine learning as it makes th…

We introduce a class of regularisable infinite dimensional principal fibre bundles which includes fibre bundles arising in gauge field theories like Yang-Mills and string theory and which generalise finite dimensional Riemannian principal fibre bundles induced by an isometric action. We show that the orbits of regulari…

New framework monitors neural network training and reveals regularisation mechanisms.

This work uncovers algorithm-dependent regularisation in diffusion models.

Despite recent advances in regularisation theory, the issue of parameter selection still remains a challenge for most applications. In a recent work the framework of statistical learning was used to approximate the optimal Tikhonov regularisation parameter from noisy data. In this work, we improve their results and ext…

The problem of adversarial examples has highlighted the need for a theory of regularisation that is general enough to apply to exotic function classes, such as universal approximators. In response, we give a very general equality result regarding the relationship between distributional robustness and regularisation, as…

New method prevents deep learning forgetting past by remembering key examples.

Novel approach finds implicit regularisation in two-player games using BEA.

Regularization preserves topological data structure in autoencoders.

Investigates gradient descent dynamics and introduces new regularisation methods.

Researchers compare different gradient methods for ridge regression, finding conjugate gradients have similar performance.

Study on GD and SGD over diagonal networks, focusing on stepsizes and regularisation.

We propose graph-dependent implicit regularisation strategies for distributed stochastic subgradient descent (Distributed SGD) for convex problems in multi-agent learning. Under the standard assumptions of convexity, Lipschitz continuity, and smoothness, we establish statistical learning rates that retain, up to logari…

This work shows how penalising bias terms in norm regularisation leads to sparse solutions.

Regularizes ML algorithms for robust multivariate analysis against distribution shifts.

Study on how noise and variation-norm regularisation help shallow ReLU networks use fewer neurons.

This work optimizes RL algorithms using entropy regularisation for continuous-time LQ problems.

Paper studies particle method for LSV model calibration, proving convergence and error bounds.

Effective regularisation of neural networks is essential to combat overfitting due to the large number of parameters involved. We present an empirical analogue to the Lipschitz constant of a feed-forward neural network, which we refer to as the maximum gain. We hypothesise that constraining the gain of a network will h…

New insights into how neural networks learn features, especially when they are very wide.

Study on neural networks with regularisation and its impact on training dynamics.

Recent work has established the equivalence between deep neural networks and Gaussian processes (GPs), resulting in so-called neural network Gaussian processes (NNGPs). The behaviour of these models depends on the initialisation of the corresponding network. In this work, we consider the impact of noise regularisation …

Machine Learning (ML) algorithms are vulnerable to poisoning attacks, where a fraction of the training data is manipulated to deliberately degrade the algorithms' performance. Optimal poisoning attacks, which can be formulated as bilevel optimisation problems, help to assess the robustness of learning algorithms in wor…

We show that many standard results of Lorentzian causality theory remain valid if the regularity of the metric is reduced to $C^{1,1}$. Our approach is based on regularisations of the metric adapted to the causal structure.

Regularising for invariance to data augmentation improves machine learning models.

Paper analyzes inclusive KL inference using Wasserstein gradient flows.

TSC uses HMC and adaptive transport maps to optimize forward KL for variational inference.

Unified understanding of three continual learning regularisation methods.

Dropout, a stochastic regularisation technique for training of neural networks, has recently been reinterpreted as a specific type of approximate inference algorithm for Bayesian neural networks. The main contribution of the reinterpretation is in providing a theoretical framework useful for analysing and extending the…

A classic setting of the stochastic K-armed bandit problem is considered in this note. In this problem it has been known that KL-UCB policy achieves the asymptotically optimal regret bound and KL-UCB+ policy empirically performs better than the KL-UCB policy although the regret bound for the original form of the KL-UCB…

New ADMM method for PARAFAC2 tensor decomposition with flexible regularization.

The paper studies $m$-positive currents and line bundles on complex manifolds.

Convolutional neural networks (CNNs) often perform well, but their stability is poorly understood. To address this problem, we consider the simple prototypical problem of signal denoising, where classical approaches such as nonlinear diffusion, wavelet-based methods and regularisation offer provable stability guarantee…

We develop a new method for regularising neural networks. We learn a probability distribution over the activations of all layers of the model and then insert imputed values into the network during training. We obtain a posterior for an arbitrary subset of activations conditioned on the remainder. This is a generalisati…

SPReD uses uncertainty to decide imitation from demonstrations.

Novel method decorrelates neurons for better deep learning model generalization.

ICCNLS models complex relationships as convex and concave components.

Improved fast rates for decision making with forward-KL regularization in contextual bandits.

Causal KL improves on existing metrics for evaluating causal models.

We present a probabilistic viewpoint to multiple kernel learning unifying well-known regularised risk approaches and recent advances in approximate Bayesian inference relaxations. The framework proposes a general objective function suitable for regression, robust regression and classification that is lower bound of the…