arXiv research

This study evaluates multi-step reinforcement learning methods in the Mountain Car environment.

DQN with model-based exploration improves learning in sparse reward environments.

Introduces LoCA regret to evaluate model-based RL methods.

The question of how to explore, i.e., take actions with uncertain outcomes to learn about possible future rewards, is a key question in reinforcement learning (RL). Here, we show a surprising result: We show that Q-learning with nonlinear Q-function and no explicit exploration (i.e., a purely greedy policy) can learn s…

This paper introduces a new scalable multi-objective deep reinforcement learning (MODRL) framework based on deep Q-networks. We develop a high-performance MODRL framework that supports both single-policy and multi-policy strategies, as well as both linear and non-linear approaches to action selection. The experimental …

Unified framework for reliable uncertainty quantification in RL.

We extend a work of Bartsch, Clapp and Puppe on the Mountain pass theorems. We consider functionals invariant with respect to infinite discrete groups satisfying a maximality condition on the finite subgroups.

The paper uses potential functions to help reinforcement learning agents learn optimal policies.

Study finds critical points in perimeter functional for fixed volume sets.

Solves time-minimizing navigation on a mountain slope using Riemann-Finsler geometry.

The geometry on a slope of a mountain is the geometry of a Finsler metric, called here the {\it slope metric}. We study the existence of globally defined slope metrics on surfaces of revolution as well as the geodesic's behavior. A comparison between Finslerian and Riemannian areas of a bounded region is also studied.

Proves uniqueness of small entropy self-expanders.

Projective simulation (PS) is a model for intelligent agents with a deliberation capacity that is based on episodic memory. The model has been shown to provide a flexible framework for constructing reinforcement-learning agents, and it allows for quantum mechanical generalization, which leads to a speed-up in deliberat…

Study examines diversification of mid-mountain ski tourism.

GPs with neural network dual kernels improve reinforcement learning performance.

Post-pandemic, work patterns shifted with fewer days in offices and a new midweek mountain.

Study improves precipitation predictions for High Mountain Asia using machine learning.

New theorem finds new minimal hypersurfaces in hyperbolic space.

This project proposes using reinforcement learning to train spiking neural networks.

Proposes a flexible MGP model for dynamic, sparse correlations.

New AI learns like neurons, generalizing from sparse rewards.

A new car following model reduces platoon level traffic errors.

A car's motion shows flat parabolic geometry.

New RL agent learns sparse rewards efficiently.

In recent years, car makers and tech companies have been racing towards self driving cars. It seems that the main parameter in this race is who will have the first car on the road. The goal of this paper is to add to the equation two additional crucial parameters. The first is standardization of safety assurance --- wh…

A kinematics of the motion of a car is reformulated in terms of the theory of gauge potentials (connection on principal bundle). E(2)-connection originates in the no-slipping contact of the car with a road.

New insights into state representations in RL help design better learning rules.

We introduce a new braid-theoretic framework with which to understand the Legendrian and transversal classification of knots, namely a Legendrian Markov Theorem without Stabilization which induces an associated transversal Markov Theorem without Stabilization. We establish the existence of a nontrivial knot-type specif…

We consider Markov Decision Problems defined over continuous state and action spaces, where an autonomous agent seeks to learn a map from its states to actions so as to maximize its long-term discounted accumulation of rewards. We address this problem by considering Bellman's optimality equation defined over action-val…

DASC combines social media and car sensors to improve disaster response.

Reduces car control labels to simplify autonomous driving.

Study on nonlinear elliptic equations with variable exponents, proving existence and multiplicity of solutions.

Ridesourcing platforms like Uber and Didi are getting more and more popular around the world. However, unauthorized ridesourcing activities taking advantages of the sharing economy can greatly impair the healthy development of this emerging industry. As the first step to regulate on-demand ride services and eliminate b…

Global diffeomorphism theorem for Fréchet spaces established.

Two methods find at least two solutions to Kazdan-Warner's problem on surfaces.

UC-SSP algorithm tackles exploration in stochastic shortest path problems without loop-free assumption.

Proposes a method for training Bayesian neural networks using synthetic data from Raman and CARS spectra.

Framework fuses RGB images and depth maps for self-driving car control.

Sparse representations improve reinforcement learning performance.

New theorem connects distant points and identical points on manifolds.

This study proposes a framework for human-like autonomous car-following planning based on deep reinforcement learning (deep RL). Historical driving data are fed into a simulation environment where an RL agent learns from trial and error interactions based on a reward function that signals how much the agent deviates fr…

While designing the state space of an MDP, it is common to include states that are transient or not reachable by any policy (e.g., in mountain car, the product space of speed and position contains configurations that are not physically reachable). This leads to defining weakly-communicating or multi-chain MDPs. In this…

RankNet forecasts car racing positions with improved accuracy and stability.

We analyze an exhaustive data-set of new-cars monthly sales. The set refers to 10 years of Spanish sales of more than 6500 different car model configurations and a total of 10M sold cars, from January 2007 to January 2017. We find that for those model configurations with a monthly market-share higher than 0.1% the sale…

Solves time-optimal navigation on slippery slopes with cross gravitational wind.

We provide sufficient conditions for the existence of a global diffeomorphism between tame Fréchet spaces. We prove a version of the Mountain Pass Theorem which is a key ingredient in the proof of the main theorem.

Reinforcement learning controls car speed for safe, efficient, and comfortable driving.

In this paper, we applied the multifractal detrended fluctuation analysis to the daily means of wind speed measured by 119 weather stations distributed over the territory of Switzerland. The analysis was focused on the inner time fluctuations of wind speed, which could be more linked with the local conditions of the hi…