Optimal switching regret for all segmentations in online convex optimisation.
problem Non-stationary online convex optimisation problems.
method Developed an efficient algorithm to achieve optimal switching regret on every possible segmentation.
result Achieved asymptotically optimal switching regret on every possible segmentation simultaneously.
Unified high-probability regret bounds for online convex optimisation with randomised gradient estimators.
problem Online convex optimisation with randomised gradient estimators for ℓ q \ell_q ℓ q -Lipschitz losses. method FTRL with randomised two-point finite-difference gradient estimators based on cone-measure sampling from ℓ r \ell_r ℓ r -spheres. result Unified high-probability regret bounds for all p , q , r ∈ [ 1 , ∞ ] p,q,r \in [1,\infty] p , q , r ∈ [ 1 , ∞ ] . Book covers tools for zeroth-order convex optimisation.
problem Zeroth-order convex optimisation.
method Cutting plane methods, interior point methods, continuous exponential weights, gradient descent, online Newton step.
result Improved existing bounds and algorithms.
Efficient algorithm for zeroth-order bandit convex optimization with bounds on regret.
problem Optimizing in unknown, noisy environments with limited information.
method Online Newton Method for bandit convex optimization, proving regret bounds.
result Regret bounds for both adversarial and stochastic settings.
The information-theoretic analysis by Russo and Van Roy (2014) in combination with minimax duality has proved a powerful tool for the analysis of online learning algorithms in full and partial information settings. In most applications there is a tantalising similarity to the classical analysis based on mirror descent.…
Paper shows equivalence between two alignment methods and introduces a new algorithm.
problem Ensuring human alignment of large language models for useful, safe, and pleasant user experience.
method Introduces IPO-MD algorithm, showing equivalence between IPO and Nash-MD methods.
result Equivalence between IPO and Nash-MD methods proven when considering online version of IPO.
Improved regret bound for adversarial bandit convex optimisation.
problem Minimizing regret in zeroth-order adversarial bandit convex optimisation.
method Identifying an improved exploratory distribution for convex functions.
result Proved minimax regret bound of O ( d 2.5 n log ( n ) ) O(d^{2.5} \sqrt{n} \log(n)) O ( d 2.5 n log ( n )) . Convex optimisation solves inverse kinematics problems more reliably.
problem Finding the best parameters of a kinematic skeleton from observed joint locations.
method Convex optimisation using semidefinite programming.
result The proposed method significantly outperforms local optimisation methods.
This paper explores optimising acquisition functions in Bayesian optimisation.
problem Optimising acquisition functions in Bayesian optimisation is challenging due to their non-convex nature.
method The authors derive compositional forms for acquisition functions and use them to recast maximisation as a compositional optimisation problem.
result The compositional approach to maximising acquisition functions shows empirical advantages across various tasks.
In this paper, we formalise order-robust optimisation as an instance of online learning minimising simple regret, and propose Vroom, a zero'th order optimisation algorithm capable of achieving vanishing regret in non-stationary environments, while recovering favorable rates under stochastic reward-generating processes.…
New method finds better arbitrage opportunities in AMMs.
problem Finding optimal arbitrage trades in multi-token AMMs.
method Closed-form solutions using convex optimisation.
result Better arbitrage opportunities than traditional methods.
Revisits online Laplace methods for neural networks, showing they are sound under certain conditions.
problem Online Laplace methods violate the Laplace approximation's critical assumption.
method Re-derives online Laplace methods, showing they target a variational bound on a mode-corrected variant of the Laplace evidence.
result Online Laplace and its mode-corrected counterpart share stationary points that satisfy the Laplace method's assumption.
Paper develops robust SGLD for solving non-convex DRO problems.
problem Solving non-convex distributionally robust optimisation problems with adversarially corrupted samples.
method Developed a Stochastic Gradient Langevin Dynamics (SGLD) algorithm with non-asymptotic convergence bounds.
result The robust SGLD estimator outperforms vanilla SGLD in terms of test accuracy.
SOBER optimizes and quadrates efficiently in parallel for diverse tasks.
problem Scalability of batch Bayesian optimization and quadrature for expensive functions.
method Reformulates batch selection as a quadrature problem, balancing exploitation and exploration.
result SOBER outperforms 11 baselines on 12 tasks.
Regression is widely used by practioners across many disciplines. We reformulate the underlying optimisation problem as a second-order conic program providing the flexibility often needed in applications. Using examples from portfolio management and quantitative trading we solve regression problems with and without con…
New method optimizes multiple points in Bayesian optimization efficiently.
problem Optimizing multiple points in expensive black-box functions.
method Reformulated BO as probability measure optimization, using convex gradient flows.
result Demonstrated effectiveness on various benchmarks compared to state-of-the-art methods.
A new framework predicts hidden Markov model regimes online.
problem Efficiently identify hidden Markov model regimes in streaming data.
method Develops a predictive-first optimisation framework for streaming HMMs, approximating the full posterior predictive distribution.
result The method provides competitive prequential performance compared to Online EM and Sequential Monte Carlo.
New methods speed up fitting for large datasets with noisy observations.
problem Fitting large datasets with Gaussian noise and known covariance.
method Two minibatch variants of extreme deconvolution, online EM algorithm, and gradient-based optimisation.
result Methods can scale to larger models and fit larger datasets faster.
A new optimisation method efficiently scales Hessian-vector products for neural networks.
problem Challenges in applying second-order quasi-Newton methods due to large Hessian and non-convexity.
method Proposes an optimisation algorithm that asymptotically uses the exact inverse Hessian with modified eigenvalues.
result Demonstrates scalability and comparable performance to other optimisation methods in neural networks.
Novel algorithms for entropic optimal transport from an optimisation perspective.
problem Solving the entropic-regularised optimal transport problem.
method Developed novel methods inspired by mirror descent, solving semi-dual problems or non-convex constrained problems over joint distributions.
result Non-asymptotic rates of convergence for the proposed methods under minimal assumptions.
Unified framework for lifted training and inversion of neural networks.
problem Challenges in gradient-based training of deep neural networks.
method Unified framework encapsulating various lifted training strategies.
result Unified framework improves training landscape and stability.
New method optimizes on curved manifolds without curvature dependence.
problem Curvature-dependent regret in online optimization on Hadamard manifolds.
method Riemannian online gradient descent for h-convex functions.
result Established O ( T ) O(\sqrt{T}) O ( T ) and O ( log ( T ) ) O(\log(T)) O ( log ( T )) regret guarantees, curvature-independent. RL models outperform traditional methods in certain market conditions.
problem Traditional portfolio management methods rely on accurate forecasts and do not incorporate specific investor preferences.
method Deep reinforcement learning with specific investor preferences incorporated into reward functions, realistic transaction costs modelled.
result RL models can significantly outperform traditional methods in upward trending markets, but not in sideways trending markets.
Online learning rbfnet improves multi-horizon returns forecasts for financial time series.
problem Nonstationarity and concept drift in financial time series.
method Combines feature representation transfer with sequential optimisation.
result Online learning rbfnet outperforms random-walk and batch learners.
New algorithm optimizes auction prices in real-time.
problem Maximizing revenue in online auctions with high frequency data.
method First real-time algorithm for online learning of monopoly prices.
result Achieves constant time and memory complexity for updates.
New approach for distributed online optimization of non-convex losses with sublinear regret.
problem Regret evaluation and consensus in distributed, multi-agent systems with non-convex losses.
method Composite regret metric and consensus-based online normalized gradient (CONGD) approach for pseudo-convex losses; offline optimization oracle for general non-convex losses.
result First sublinear regret bound for general distributed online non-convex learning.
The inverse Ising problem seeks to reconstruct the parameters of an Ising Hamiltonian on the basis of spin configurations sampled from the Boltzmann measure. Over the last decade, many applications of the inverse Ising problem have arisen, driven by the advent of large-scale data across different scientific disciplines…
New research challenges the independence assumption in neurosymbolic learning, leading to overconfident predictions and unrepresentable uncertainty.
problem The independence assumption in neurosymbolic learning systems can lead to overconfident predictions and hinder uncertainty quantification.
method The study proves the limitations of the independence assumption and introduces new loss functions that are non-convex and difficult to optimise.
result Neurosymbolic learning systems using the independence assumption are prone to overconfidence and cannot represent uncertainty over multiple valid options.
A new optimisation framework for neural networks without hyperparameters.
problem Lack of explicit architectural information in optimisation frameworks.
method Extends mirror descent to account for neural architecture, transforming a Bregman divergence.
result Automatic gradient descent: a first-order optimiser without hyperparameters.
Improved online learning for hidden-convex losses achieves optimal regret.
problem Adversarial online learning with nonconvex losses that become convex after reparameterization.
method Algorithmic equivalence between OGD and OMD on convex losses, with Hessian compatibility condition.
result OGD achieves O ( T ) \mathcal{O}(\sqrt{T}) O ( T ) regret for hidden-convex losses, matching optimal rate. Boosting improves online decision-making for large expert sets.
problem Online convex optimization with many experts is infeasible.
method Generalizes online boosting to online convex optimization and bandit linear optimization settings.
result Near-optimal regret guarantees for various feedback models.
New algorithm for online convex minimization over integer lattice.
problem Online decision-making with nonlinear combinatorial objectives.
method Introduces online L a t u r a l \mathrm{L}^{
atural} L a t u r a l -convex minimization and proposes efficient algorithms. result Tight regret bound for full information setting algorithm.
New algorithm SFHC achieves near-optimal costs with predictions for non-convex optimization.
problem Online optimization with non-convex hitting costs and movement costs.
method Synchronized Fixed Horizon Control (SFHC) algorithm with conditions on hitting and movement costs.
result Synchronized Fixed Horizon Control (SFHC) achieves a 1 + O ( 1 / w ) 1+O(1/w) 1 + O ( 1/ w ) competitive ratio for near-optimal costs. Higher conservative training increases reward-hacking in reasoning models.
problem Reward hacking during online adaptation in reasoning models.
method Conservative offline training with varying levels of conservatism (β) was applied to a Qwen3-14B policy, and online adaptation was measured against a reward ensemble.
result Higher conservatism (β) increases reward-hacking damage, measured by the Goodhart gap and AUGC.
Extends boosting to multiclass online agnostic classification.
problem Online multiclass classification with weak learners.
method Reduces multiclass online agnostic boosting to online convex optimization.
result First boosting algorithm for online agnostic multiclass classification.
New algorithm reduces regret in stochastic bandit convex optimization.
problem Optimizing decisions in uncertain environments with convex losses.
method Introduces a second-order method for zeroth-order stochastic convex bandits.
result Regret bound of ( 1 + r / d ) [ d 1.5 n + d 3 ] p o l y l o g ( n , d , r ) (1 + r/d)[d^{1.5} \sqrt{n} + d^3] polylog(n, d, r) ( 1 + r / d ) [ d 1.5 n + d 3 ] p o l y l o g ( n , d , r ) . New algorithm for online learning in episodic MDPs with convex objectives.
problem Online episodic convex reinforcement learning.
method Online mirror descent algorithm with varying constraint sets and exploration bonus.
result Near-optimal regret bounds for online CURL without prior knowledge of transition function.
New control methods for systems with adversarial perturbations.
problem Control systems with adversarial noise.
method Online convex optimization and convex relaxations.
result Low regret policies against adversarial perturbations.
Optimal algorithms for online convex optimization with random order.
problem Online convex optimization with random order and non-convex loss functions.
method Stochastic gradient descent and algorithmic stability analysis.
result Achieves optimal bounds and significantly outperforms previous methods.
Boosts weak online learners to strong ones with sublinear regret.
problem Online learning agnostic setting without strong guarantees.
method Reduction to online convex optimization, boosting via marginally-better-than-trivial regret guarantees.
result First agnostic online boosting algorithm with sublinear regret.
New algorithms optimize non-smooth, non-convex objectives with improved complexity.
problem Optimizing non-smooth, non-convex stochastic objectives.
method Reduction to online learning, applying optimistic online learning techniques.
result Improved complexity for finding ( δ , ε ) (δ,ε) ( δ , ε ) -stationary points. Optimal hidden-target learning for online inventory optimization on general convex sets.
problem Online inventory optimization (OIO) on arbitrary bounded convex capacity sets.
method Maintaining a hidden target and projecting it onto the feasible order-up-to set.
result The method improves the best known regret guarantee for OIO on general convex sets from inverse to inverse-square-root dependence on the common-demand probability.
Optimal control in changing systems without strong convexity assumptions.
problem Adversarial changes in convex costs for unknown linear systems.
method Non-convex lower confidence bounds and computationally-efficient regret minimization.
result Achieves T \smash{\sqrt{T}} T -regret rate, optimal compared to best stabilizing controller. New framework uses tempered optimism to handle imperfect experts in online learning.
problem Challenges of implicit optimism in practical online learning environments.
method Introduces tempered optimism as a framework for online non-convex learning, modifies existing algorithms.
result Demonstrates tempered optimism as a fruitful paradigm for online non-convex learning.
New framework captures long-term decision dependence in online learning.
problem Long-term dependence on past decisions in online learning.
method Introduces Online Convex Optimization with Unbounded Memory (OCO-UMB) and p p p -effective memory capacity. result Proves O ( H p T ) O(\sqrt{H_p T}) O ( H p T ) upper bound on policy regret and matching lower bound. Online optimization has been a successful framework for solving large-scale problems under computational constraints and partial information. Current methods for online convex optimization require either a projection or exact gradient computation at each step, both of which can be prohibitively expensive for large-scal…
Paper tackles online optimization with memory and competitive control.
problem Minimizing hitting and switching costs in online optimization problems.
method Optimistic Regularized Online Balanced Descent algorithm.
result Achieves a constant, dimension-free competitive ratio.
Paper extends meta-learning framework to non-convex settings with improved performance.
problem Learning from past tasks for faster future tasks in a sequential setting.
method Generalized online meta-learning framework to non-convex settings, introduced local regret as performance measure.
result The framework achieves logarithmic local regret and robustness to hyperparameter initialization.