arXiv research

A new model-free algorithm achieves near-optimal regret for infinite-horizon MDPs.

Improved model-free RL algorithm with reduced sample complexity.

Improved model-free reinforcement learning with decision-estimation coefficient.

RandQL is a new model-free algorithm for MDPs with a novel learning rate randomization approach.

UCB-Advantage learns MDPs with $O(\sqrt{H^2SAT})$ regret.

Improved bounds on the copula of a bivariate random vector are computed when partial information is available, such as the values of the copula on a given subset of $[0,1]^2$, or the value of a functional of the copula, monotone with respect to the concordance order. These results are then used to compute model-free bo…

Deep learning for financial derivatives pricing and hedging.

We study the sample complexity of model-based reinforcement learning (henceforth RL) in general contextual decision processes that require strategic exploration to find a near-optimal policy. We design new algorithms for RL with a generic model class and analyze their statistical properties. Our algorithms have sample …

New algorithm tackles non-stationary RL with near-optimal regret bounds.

Paper develops efficient algorithms for zero-sum Markov games with general function classes.

In a model free discrete time financial market, we prove the superhedging duality theorem, where trading is allowed with dynamic and semi-static strategies. We also show that the initial cost of the cheapest portfolio that dominates a contingent claim on every possible path $ω\in Ω$, might be strictly greater than the …

A new method for matrix completion with model-free weights.

The paper develops bounds for multi-asset derivatives using option prices.

New algorithms reduce regret in both stochastic and deterministic environments.

A new method stabilizes deep reinforcement learning by using QGraphs to retain replay memory information.

Unified Latent Dynamics unifies model-free and model-based reinforcement learning.

Paper proposes a model-free algorithm for CMDPs with long-term constraints, achieving optimal regret bounds.

New method refines model-free evaluation of complex machine learning models.

Study compares model-free valuation to actual financial outcomes, finds it slightly conservative.

Researchers find a way to price American options without relying on specific asset price models.

ZDPG learns model-free policies without critics, improving on PG.

Paper proves efficiency of MARL with transformers, addressing agent complexity.

Algorithm learns NE in imperfect information games with imperfect feedback.

We discuss the possibility of obtaining model-free bounds on volatility derivatives, given present market data in the form of a calibrated local volatility model. A counter-example to a wide-spread conjecture is given.

New model-free DR-RL algorithm with finite sample complexity.

New RL algorithm achieves sublinear regret and constraint violation without simulators.

We prove the existence and uniqueness of solutions of SDEs with Lipschitz coefficients, driven by continuous, model-free martingales. The main tool in our reasoning is Picard's iterative procedure and a model-free version of the Burkholder-Davis-Gundy inequality for integrals driven by model-free, continuous martingale…

Paper develops efficient RL algorithm for general value function approximation.

Unified reinforcement learning methods using hybrid inference.

New model-free algorithm achieves similar LQR regret guarantees.

Paper improves neural network robustness analysis for safety-critical systems.

FORK improves model-free reinforcement learning performance.

Study dynamic trading in options to improve price bounds for exotic derivatives.

We derive upper and lower bounds on the expectation of $f(\mathbf{S})$ under dependence uncertainty, i.e. when the marginal distributions of the random vector $\mathbf{S}=(S_1,\dots,S_d)$ are known but their dependence structure is partially unknown. We solve the problem by providing improved \FH bounds on the copula o…

New algorithms reduce risk in reinforcement learning with provable regret bounds.

Model-free reinforcement learning is known to be memory and computation efficient and more amendable to large scale problems. In this paper, two model-free algorithms are introduced for learning infinite-horizon average-reward Markov Decision Processes (MDPs). The first algorithm reduces the problem to the discounted-r…

Fine-grained gap-dependent regret bounds for reinforcement learning.

We study specific nonlinear transformations of the Black-Scholes implied volatility to show remarkable properties of the volatility surface. Model-free bounds on the implied volatility skew are given. Pricing formulas for the European options which are written in terms of the implied volatility are given. In particular…

Model-free deep reinforcement learning has been shown to exhibit good performance in domains ranging from video games to simulated robotic manipulation and locomotion. However, model-free methods are known to perform poorly when the interaction time with the environment is limited, as is the case for most real-world ro…

New RL algorithm reduces sample complexity for optimal learning.

Improved bounds for multi-asset options using deep learning and market prices.

Efficient exploration for automatic subgoal discovery is a challenging problem in Hierarchical Reinforcement Learning (HRL). In this paper, we show that intrinsic motivation learning increases the efficiency of exploration, leading to successful subgoal discovery. We introduce a model-free subgoal discovery method base…

New hybrid RL algorithm outperforms model-free and model-based methods.

VMBPO optimizes model and policy jointly using variational lower-bound.

Simple object representations improve model-free RL performance.

Paper proposes FedQ-Advantage for federated Q-learning with near-optimal regret and low communication cost.

Develops model-free methods for event history analysis and efficient covariate adjustment.

New algorithms for model selection in off-policy evaluation of reinforcement learning.