arXiv research

Neural networks' weights don't converge to stationary points but training loss stabilizes.

New method stabilizes FQE by reweighting Bellman targets.

This paper refines the weighted strategy for non-stationary parametric bandits and MDPs, improving regret bounds.

This paper refines the weighted strategy for non-stationary parametric bandits, improving regret bounds.

In this paper we generalize examples of Hamiltonian stationary Lagrangian submanifolds constructed by Lee and Wang in $\mathbb{C}^m$ to toric almost Calabi-Yau manifolds. We construct examples of weighted Hamiltonian stationary Lagrangian submanifolds in toric almost Calabi-Yau manifolds and solutions of generalized La…

Optimistic algorithm reduces regret in non-stationary linear MDPs.

Novel Bayesian approach for non-stationary linear contextual bandits.

Cautious Weight Decay modifies weight decay for better optimization.

In this paper we consider a class of weighted-volume preserving curvature flows acting on hypersurfaces that are trapped within two parallel hyperplanes and satisfy an orthogonal boundary condition. In the author's thesis the stability of cylinders under the flows was considered; it was found that they are stable provi…

New method improves stability of soft FQI for offline RL.

Framework models supervised learning in non-stationary data.

We consider a stochastic linear bandit model in which the available actions correspond to arbitrary context vectors whose associated rewards follow a non-stationary linear regression model. In this setting, the unknown regression parameter is allowed to vary in time. To address this problem, we propose D-LinUCB, a nove…

New algorithm reduces dynamic regret in non-stationary dueling bandits using a weighted Borda score.

This paper introduces the Partition Tree Weighting technique, an efficient meta-algorithm for piecewise stationary sources. The technique works by performing Bayesian model averaging over a large class of possible partitions of the data into locally stationary segments. It uses a prior, closely related to the Context T…

Study tackles non-stationary bandit convex optimization with new algorithms.

New method detects intrinsic cross-correlations in non-stationary time series affected by common factors.

Standard kernels such as Matérn or RBF kernels only encode simple monotonic dependencies within the input space. Spectral mixture kernels have been proposed as general-purpose, flexible kernels for learning and discovering more complicated patterns in the data. Spectral mixture kernels have recently been generalized in…

Study neural architectures on learned latent graphs using Schrödinger dynamics.

We consider the problem of learning over non-stationary ranking streams. The rankings can be interpreted as the preferences of a population and the non-stationarity means that the distribution of preferences changes over time. Our goal is to learn, in an online manner, the current distribution of rankings. The bottlene…

Innovative game theory approach optimizes survival analysis metrics.

Study on consistency of ML methods for moving objects in non-stationary environments.

New algorithm reduces online learning regret by exploiting historical invariances.

New GLB algorithm handles non-stationary data with forgetting.

The study connects minimal and maximal surfaces in 3D and 3-L space.

New Hermite series estimator for Spearman rank correlation in non-stationary data.

New algorithm clusters trajectories from multiple Markov chains with near-optimal error.

Paper develops sparse learning for heavy-tailed time series with locally stationary dynamics.

The paper analyzes prediction error in nonstationary settings using weighted risk minimization.

A new metric, Weighted Regret, unifies FDR and power evaluation in online multiple testing.

Gradient descent with random weights in linear regression analyzed for various noise types.

We present a modelling framework for the investigation of prototype-based classifiers in non-stationary environments. Specifically, we study Learning Vector Quantization (LVQ) systems trained from a stream of high-dimensional, clustered data.We consider standard winner-takes-all updates known as LVQ1. Statistical prope…

Adaptive beamforming collapses in highly non-stationary environments, but the Universal Switching Beamformer resolves this by dynamically adjusting memory length.

FredNormer improves time series forecasting by adapting to frequency domain patterns.

Researchers have constantly asked whether stock returns can be predicted by some macroeconomic data. However, it is known that macroeconomic data may exhibit nonstationarity and/or heavy tails, which complicates existing testing procedures for predictability. In this paper we propose novel empirical likelihood methods …

This paper studies the landscape of empirical risk of deep neural networks by theoretically analyzing its convergence behavior to the population risk as well as its stationary points and properties. For an $l$-layer linear neural network, we prove its empirical risk uniformly converges to its population risk at the rat…

Paper introduces novel Bandit algorithms for non-stationary environments in finance.

Estimates and infers multi-stage stationary treatment policies with variable selection.

ABO extends RLS for online learning in non-stationary time-series, improving accuracy and speed.

A new memory system handles non-stationary environments by self-sizing and retaining memories.

New algorithm provably converges to second-order stationary points in NMF.

Graphs are a central tool in machine learning and information processing as they allow to conveniently capture the structure of complex datasets. In this context, it is of high importance to develop flexible models of signals defined over graphs or networks. In this paper, we generalize the traditional concept of wide …

Algorithm learns stochastic system dynamics from data.

KQT-EWMA monitors multivariate data streams online with flexible and practical change detection.

Paper introduces a new method for improving reinforcement learning performance using transfer learning.

New covariance estimator for financial portfolios.

Introduces a new theoretical framework for exponential smoothing.

We consider the teacher-student setting of learning shallow neural networks with quadratic activations and planted weight matrix $W^*\in\mathbb{R}^{m\times d}$, where $m$ is the width of the hidden layer and $d\le m$ is the data dimension. We study the optimization landscape associated with the empirical and the popula…

MELO predicts electricity loads by adapting to shifts without external indicators.