NM-PPG optimizes adaptive feature acquisition in POMDPs for better predictions.
problem Optimizing adaptive feature acquisition in prediction problems with costly features.
method Non-myopic pathwise policy gradients (NM-PPG) with continuous relaxation and straight-through rollout.
result NM-PPG outperforms state-of-the-art AFA methods on synthetic and real-world datasets.
This work proposes using zero-variance control variates to reduce variance in pathwise gradient estimators for variational inference.
problem Pathwise gradient estimators in variational inference have high variance, leading to inefficient optimization.
method Apply zero-variance control variates to pathwise gradient estimators.
result Zero-variance control variates can significantly reduce the variance of pathwise gradient estimators without requiring complex assumptions.
A new reinforcement learning method uses model derivatives to improve policy optimization.
problem Improving sample efficiency and performance in model-based reinforcement learning.
method Constructs an actor-critic algorithm that uses the pathwise derivative of the learned model and policy.
result Consistently more sample efficient and matches model-free algorithms' asymptotic performance.
New algorithm reduces variance in Monte Carlo simulations using deep neural networks and policy gradients.
problem Reducing variance in Monte Carlo simulations for estimating function values.
method Optimal correlation search using deep neural networks and policy gradients.
result Optimal correlation function reduces variance by approximating and calibrating policy.
Previously, the exploding gradient problem has been explained to be central in deep learning and model-based reinforcement learning, because it causes numerical issues and instability in optimization. Our experiments in model-based reinforcement learning imply that the problem is not just a numerical issue, but it may …
We observe that gradients computed via the reparameterization trick are in direct correspondence with solutions of the transport equation in the formalism of optimal transport. We use this perspective to compute (approximate) pathwise gradients for probability distributions not directly amenable to the reparameterizati…
Second-order optimization speeds up deep hedging for complex options.
problem Hedging exotic options with market frictions in realistic markets.
method Second-order optimization scheme leveraging pathwise differentiability and Kronecker-factoring.
result Our method optimizes the policy in 1/4 the steps of standard optimization.
We exploit the link between the transport equation and derivatives of expectations to construct efficient pathwise gradient estimators for multivariate distributions. We focus on two main threads. First, we use null solutions of the transport equation to construct adaptive control variates that can be used to construct…
Hybrid Policy Optimization tackles reinforcement learning in hybrid spaces, improving performance over PPO.
problem Credit assignment issues and biased gradients in hybrid discrete-continuous action spaces.
method Mixed gradient estimator combining pathwise and score-function gradients, reformulating problems in hybrid form.
result HPO substantially outperforms PPO on inventory control and switched systems, with performance gaps increasing with continuous action dimension.
A scalable method for BED with implicit models using approximate gradients.
problem Efficiently estimating posterior distribution and maximizing MI for implicit models.
method Stochastic approximate gradient ascent with smoothed variational MI estimator.
result Significantly improves scalability of BED in high-dimensional problems.
A new method in finance without probabilities or integrals.
problem Creating a model-free approach to continuous-time finance.
method Pathwise approach using causal functional calculus and transition principle of Isaacs.
result A fully non-linear path-dependent equation characterizes optimal solutions.
We propose a stochastic approximation (SA) based method with randomization of samples for policy evaluation using the least squares temporal difference (LSTD) algorithm. Our proposed scheme is equivalent to running regular temporal difference learning with linear function approximation, albeit with samples picked unifo…
The paper speeds up hyperparameter optimisation in Gaussian processes.
problem Scaling hyperparameter optimisation to large datasets.
method Improvements to linear system solvers (pathwise gradient, warm starting, early stopping).
result Speed-ups of up to 72x and residual norm decreases of up to 7x.
The paper optimizes bridge-type estimators for sparse models using pathwise methods.
problem Sparse parametric models with adaptive coefficients and multiple penalties.
method Pathwise optimization with accelerated proximal gradient descent and blockwise alternating optimization.
result Efficient computation of the full solution path for adaptive bridge estimators.
We consider a strictly pathwise setting for Delta hedging exotic options, based on Föllmer's pathwise Itō calculus. Price trajectories are d-dimensional continuous functions whose pathwise quadratic variations and covariations are determined by a given local volatility matrix. The existence of Delta hedging strategie…
Improves BED scalability for implicit models.
problem Designing experiments for implicit models with intractable data distributions.
method Hybrid gradient approach combining variational MI estimator, ES, and SGA.
result Significantly improves scalability of BED for implicit models.
Pathwise uniqueness shown for specific stochastic equations.
problem Stochastic Volterra equations with singular kernels and Hölder coefficients.
method Established pathwise uniqueness through Hölder continuity of coefficients.
result Pathwise uniqueness and existence of unique strong solutions.
This work extends ME-RL using diffusion models to sample optimal policies.
problem Sampling from the optimal policy trajectory distribution in ME-RL.
method Introducing Diffusion-Augmented Markov Decision Processes (DA-MDPs) to minimize reverse KL divergence.
result DA-MDPs enable seamless integration into various ME-RL methods and outperform baselines.
Develops pathwise analysis for log-optimal portfolios using rough paths theory.
problem Analyzing stability and approximation of log-optimal portfolios.
method Pathwise approach based on càdlàg rough paths theory.
result Establishes pathwise stability and error estimates for log-optimal portfolios.
Develops portfolio theory without probabilistic analysis, focusing on pathwise decomposition.
problem Ensuring market viability without probabilistic assumptions.
method Uses pathwise decomposition and trend extractors to replace semimartingale decomposition.
result Growth-numéraire and viability equivalences are similar but not identical in pathwise setting.
A new approach to continuous-time universal portfolios using pathwise Itô calculus.
problem Continuous-time version of Cover's universal portfolio strategies.
method Pathwise Itô calculus approach to establish existence and properties of universal portfolio strategies.
result The universal portfolio strategy's portfolio value process is the average of all values of constant rebalanced strategies.
The pathwise coordinate optimization is one of the most important computational frameworks for high dimensional convex and nonconvex sparse learning problems. It differs from the classical coordinate optimization algorithms in three salient features: {\it warm start initialization}, {\it active set updating}, and {\it …
Unified market making controls risk, arbitrage, and volatility surfaces.
problem Market making risk, arbitrage, and volatility surface consistency.
method Constrained RL and stochastic control for risk-sensitive execution and hedging.
result Agent achieves positive P&L with zero calendar and butterfly violations.
We use pathwise Itô calculus to prove two strictly pathwise versions of the master formula in Fernholz' stochastic portfolio theory. Our first version is set within the framework of Föllmer's pathwise Itô calculus and works for portfolios generated from functions that may depend on the current states of the market port…
Efficient inference for adaptive data with directional stability condition.
problem Efficient inference on scalar targets after adaptive data collection.
method Introduces directional stability, a weaker condition than i.i.d. data, and shows asymptotic normality and efficiency of estimators.
result Estimators remain asymptotically normal and semiparametrically efficient under directional stability.
This paper develops a mathematical framework for the analysis of continuous-time trading strategies which, in contrast to the classical setting of continuous-time mathematical finance, does not rely on stochastic integrals or other probabilistic notions. Our purely analytic framework allows for the derivation of a path…
This paper simplifies hedge ratios in financial models using pathwise algorithmic differentiation.
problem Expensive and unstable computation of hedge ratios from pathwise sensitivities.
method Develops reduced stochastic hedge ratios of the form φ_j^r = Σ_j^r ξ_j^q X_q, retaining sensitivity tensor through empirical averages.
result Two coefficient criteria are introduced to minimize pathwise residuals and satisfy moment equations.
The Monte Carlo pathwise sensitivities approach is well established for smooth payoff functions. In this work, we present a new Monte Carlo algorithm that is able to calculate the pathwise sensitivities for discontinuous payoff functions. Our main tool is to combine the one-step survival idea of Glasserman and Staum wi…
This paper gives several simple constructions of the pathwise Ito integral ∫0tφdω for an integrand φ and a price path ω as integrator, with φ and ω satisfying various topological and analytical conditions. The definitions are purely pathwise in that neither φ nor ω are assumed to be paths of stochast…
Off-policy stochastic actor-critic methods rely on approximating the stochastic policy gradient in order to derive an optimal policy. One may also derive the optimal policy by approximating the action-value gradient. The use of action-value gradients is desirable as policy improvement occurs along the direction of stee…
This dissertation advances scalable Gaussian processes using iterative methods and pathwise conditioning.
problem The classical Gaussian process formulation is not scalable for large datasets and modern hardware.
method Combining iterative methods and pathwise conditioning to improve scalability.
result Significantly reduced memory requirements and facilitated application to larger datasets.
This work introduces efficient sampling methods for Gaussian processes by focusing on pathwise conditioning.
problem Intractable mathematical expressions in Gaussian process posteriors limit practical applications.
method Investigates a pathwise interpretation of conditioning to derive efficient sampling methods.
result Derives a general family of approximations that allow for efficient sampling of Gaussian process posteriors.
MuRiT efficiently computes multi-parameter persistence barcodes.
problem Efficient computation of multi-parameter persistent homology.
method Vietoris-Rips transformation to reduce multi-parameter to single-parameter computation.
result MuRiT computes pathwise persistence barcodes for multi-filtered flag complexes.
The goal of policy gradient approaches is to find a policy in a given class of policies which maximizes the expected return. Given a differentiable model of the policy, we want to apply a gradient-ascent technique to reach a local optimum. We mainly use gradient ascent, because it is theoretically well researched. The …
Faster policy learning via continuous-time gradients.
problem Efficiently estimating policy gradients for continuous-time systems.
method Approximating continuous-time gradients directly, using adaptive discretization.
result More efficient policy gradient estimator leads to faster learning.
STORM-PG uses momentum for faster policy gradient updates.
problem Improving policy gradient methods for reinforcement learning.
method Introduces STORM-PG, a SARAH-based algorithm with exponential moving average.
result Achieves O(1/ε3) sample complexity, matching best-known rate. We study the use of the multilevel Monte Carlo technique in the context of the calculation of Greeks. The pathwise sensitivity analysis differentiates the path evolution and reduces the payoff's smoothness. This leads to new challenges: the inapplicability of pathwise sensitivities to non-Lipschitz payoffs often makes …
Optimizes Thompson sampling policies using policy gradient methods.
problem Improving Thompson sampling in bandit problems.
method Applies policy gradient algorithms to optimize Thompson sampling policies.
result Direct policy search on Thompson sampling improves performance.
Policy gradient methods achieve linear convergence in simple MDPs.
problem Analyzing convergence rates of policy gradient methods in finite MDPs.
method Connections with policy iteration to show linear convergence with large step-sizes.
result Policy gradient methods succeed with large step-sizes and achieve linear rate of convergence.
We investigate whether it is possible to formulate option pricing and hedging models without using probability. We present a model that is consistent with two notions of volatility: a historical volatility consistent with statistical analysis, and an implied volatility consistent with options priced with the model. The…
DBQPG improves policy gradient estimation with fewer samples.
problem Accurate policy gradient estimation with limited samples.
method Deep Bayesian Quadrature Policy Gradient (DBQPG).
result DBQPG provides more accurate and less variable gradient estimates.
New k-step policy gradient method avoids local optima in restricted policy classes.
problem Suboptimal local optima in policy gradient methods for restricted policy classes.
method Proposes a k-step policy gradient method to escape myopic local optima. result The method converges to near optimal solutions exponentially close to the optimal deterministic policy.
Due to the high variance of policy gradients, on-policy optimization algorithms are plagued with low sample efficiency. In this work, we propose Augment-Reinforce-Merge (ARM) policy gradient estimator as an unbiased low-variance alternative to previous baseline estimators on tasks with binary action space, inspired by …
Policy gradient methods are widely used for control in reinforcement learning, particularly for the continuous action setting. There have been a host of theoretically sound algorithms proposed for the on-policy setting, due to the existence of the policy gradient theorem which provides a simplified form for the gradien…
The paper interprets policy-gradient algorithms using continuation theory.
problem Optimizing nonconvex functions in reinforcement learning.
method Formulates policy optimization as optimization by continuation, interprets policy-gradient algorithms as implicitly optimizing deterministic policies.
result Exploration in policy-gradient algorithms is seen as computing a continuation of the return of the policy.
Efficient estimators for smooth Hilbert-valued parameters with theoretical guarantees.
problem Estimating smooth Hilbert-valued parameters with theoretical guarantees.
method Pathwise differentiable Hilbert-valued parameters, efficient influence functions, regularized one-step estimators.
result Theoretical guarantees for efficient estimators even when nuisance functions are arbitrary.
Large deviations theory applied to policy gradient methods.
problem Understanding convergence of policy gradient methods in reinforcement learning.
method Large deviation rate function and contraction principle from large deviations theory.
result Convergence properties of policy gradient methods can be extended to various policy parametrizations.
We propose expected policy gradients (EPG), which unify stochastic policy gradients (SPG) and deterministic policy gradients (DPG) for reinforcement learning. Inspired by expected sarsa, EPG integrates across the action when estimating the gradient, instead of relying only on the action in the sampled trajectory. We es…