Network Lasso optimizes ride sharing predictions for real-time requests.
problem Real-time ride sharing prediction with adaptive and scalable techniques.
method Network Lasso optimization framework using ADMM optimization.
result Minimized mean squared error in prediction, demonstrating effectiveness.
Paper tackles ride-sharing user experience enhancement with weakly supervised learning.
problem Compound weakly supervised learning problem in ride-sharing comment data.
method CWSL method with instance reweighting, robust criteria, and alternating optimization.
result Effectiveness validated on Didi ride-sharing comment data.
Study predicts individual treatment effects in ride-sharing competitions.
problem Understanding how team competitions affect individual drivers' outcomes.
method Analyzed data from 500 competitions, built machine learning models.
result Reduced out-sample prediction error by over 24%.
A risk-sensitive approach to inverse reinforcement learning.
problem Risk-sensitive reinforcement learning in Markov decision processes.
method Gradient-based algorithm minimizing a loss function defined on observed behavior.
result Demonstrated performance on Grid World and ride-sharing examples.
Deep RL tackles fleet management and dispatching for ride-sharing platforms.
problem Optimizing dispatching and repositioning of drivers in ride-sharing platforms.
method Deep reinforcement learning approach treating drivers as a central system agent.
result Centralized decision-making improves overall fleet efficiency.
Develops efficient algorithms for summarizing large datasets.
problem Summarizing large datasets efficiently.
method Two-stage submodular approach for linear-time optimization.
result Achieves nearly-optimal solutions for summarizing data.
Proposes LC-ST-FCN for better ride-sourcing demand forecasting.
problem Local statistical differences in ride-sourcing demand across a city.
method LC-ST-FCN framework combining 3D and 2D convolutions, locally connected layers.
result Significant improvements in demand forecasting compared to baselines.
This chapter reviews statistical tools for reinforcement learning.
problem Applying RL algorithms in healthcare and ride-sharing platforms.
method Statistical inference tools for RL, including hypothesis testing and confidence interval construction.
result Highlighting the value of statistical inference in RL for both communities.
DN estimator mitigates network interference in experiments.
problem Network interference biases naive experiment designs.
method Differences-in-Neighbors (DN) estimator designed to mitigate interference.
result DN achieves bias second order in interference effect, with exponentially smaller variance.
The paper tackles ride-hailing fleet repositioning with a calibrated demand approach.
problem Repositioning idle supply before future demand is observed in ride-hailing.
method A predict-then-optimize approach using calibrated demand regimes, a similarity gate, and spatial queue-regret decomposition.
result The spatial gate reduces mean wait time to 82.3s compared to 85.3s for a hand-tuned similarity gate and 85.8s for a distributional-only baseline.
Predicts destinations and routes from partial trajectory data.
problem Predicting destinations and routes from partial trajectory data for applications like parking suggestions and ride-sharing.
method Three-step procedure: k-d tree-based space discretization, recurrent neural network for destination prediction, and route calculation.
result Best models predict destinations with a mean error of 1.3 km and 1.43 km.
A new framework uses multi-agent reinforcement learning for evaluating policies in two-sided markets.
problem Evaluating the effects of different policies in two-sided markets with spatial and temporal interference.
method Introduces a multi-agent reinforcement learning (MARL) framework to address policy evaluation challenges in large-scale fleet management.
result Proposes novel estimators for mean outcomes under different products that are consistent despite high-dimensionality.
A new ride-hailing subsidy system uses deep causal networks to estimate consumer elasticity.
problem Estimating consumer elasticity with subsidies in ride-hailing industry.
method Introduces a consumer subsidizing system using deep causal networks to address confounding effects.
result Effective in estimating the uplift effect of subsidies without confounding.
The paper investigates machine learning methods for ride-hailing demand forecasting.
problem Characterizing and forecasting spatio-temporal demand in ride-hailing services.
method Various machine learning models (single decision tree, bagged decision trees, random forest, boosted decision trees, artificial neural network) were compared using statistical metrics.
result Boosted decision trees provided the best prediction accuracy (RMSE=16.41) among the tested models.
Study on optimal bubble riding with price-dependent entry times in a mean field game model.
problem Optimal bubble riding with price-dependent entry times.
method Mean field game of controls with common noise and random entry time, existence result obtained through discretization and limit analysis.
result Existence of equilibrium in the mean field game model.
Predict bike flow at station-level with multi-graph CNNs.
problem Fine-grained bike flow prediction for station-level management.
method Multi-graph convolutional neural network model.
result Reduces 25.1% and 17.0% prediction error in New York City and Chicago.
FastSecAgg improves federated learning security and efficiency.
problem Privacy leakage in federated learning due to model parameter sharing.
method Introduces FastSecAgg, a secure aggregation protocol with FFT-based multi-secret sharing (FastShare).
result Efficient in computation and communication, robust to client dropouts.
Study examines ridesourcing patterns in Chicago using K-prototypes segmentation.
problem Limited data on ridesourcing trips hinders research on urban mobility.
method Utilized K-prototypes algorithm for mixed data types segmentation.
result Six ridesourcing user segments identified based on various factors.
ChoiceRank learns network edge probabilities from node traffic data.
problem Learning edge transition probabilities from node-level traffic data.
method Preference learning model based on Luce's axiom, iterative algorithm.
result Successfully recovers edge transition probabilities from marginal node traffic data.
Efficiently projects points onto polytopes, especially useful in web-scale applications.
problem Efficiently projecting points onto polytopes in large-scale applications.
method Developed a vertex-oriented incremental algorithm for polytope projection, tailored for simplex and unit-box cut polytopes.
result Majority of projections lie on vertices of polytopes, leading to significant performance improvements.
The paper analyzes data markets with multiple data aggregators, showing non-uniqueness of equilibria and social inefficiency.
problem Non-uniqueness of equilibria and social inefficiency in data markets with multiple data aggregators.
method Characterization of generalized Nash equilibria and analysis of necessary and sufficient conditions for social inefficiency.
result There are either infinitely many or no generalized Nash equilibria, leading to social inefficiency.
A new method improves ridesharing efficiency using QMIX.
problem Improving ridesharing dispatch efficiency with complex environments.
method QMIX for centralized training with decentralized execution.
result QMIX outperforms IDQN in various scenarios.
Detects ridesourcing cars from ordinary vehicles using transfer learning.
problem Unauthorized ridesourcing activities impairing the industry.
method Two-stage transfer learning framework: RF for initial labeling, CNN for final detection.
result Achieves similar accuracy to supervised learning methods without labeled data.
We study Exo-MDPs to reduce sample complexity in reinforcement learning.
problem Reducing sample complexity in reinforcement learning for structured MDPs.
method Introducing Exo-MDPs and proving structural equivalence to linear mixture MDPs, establishing regret bounds.
result Proved O ( H 3 / 2 d K ) O(H^{3/2}d\sqrt{K}) O ( H 3/2 d K ) regret bound for Exo-MDPs, matching lower bounds. Method uses pseudo-samples to improve RCT data in ride-hailing pricing studies.
problem Small and biased RCT data leads to significant bias when generalizing to broader user base.
method Pseudo-sample matching to expand and match RCT data with observational data.
result 0.41% improvement in profit through pseudo-sample matching.
Proposes a graph neural network for efficient multi-agent routing.
problem Routing multiple agents in complex, sparsely connected graphs with dynamic traffic.
method Value Iteration Network based on graph neural networks for online coordination.
result Significantly outperforms traditional methods in terms of cost and runtime.
Agents can achieve minimal regret by observing others, reducing overall decision-making cost.
problem Exploration cost and unfairness of free-riding in multi-agent decision-making.
method Analyzes multi-armed bandit problem with multiple decision makers, showing free riders can achieve minimal regret under certain conditions.
result Free riders can achieve O ( 1 ) O(1) O ( 1 ) regret by observing self-reliant agents' strategies and arm pulls, without needing to observe rewards. Paper tackles NP-hard multi-agent planning with reinforcement learning.
problem Solving NP-hard multi-agent, multi-task planning problems with time-dependent rewards.
method Developed a reinforcement learning framework using mean-field inference and auction-based selection.
result Achieved near-optimality and transferability in solving MRRC and IPMS problems.
We propose a new heavy-tailed distribution --- Gaussian-Chain (GC) distribution, which is inspirited by the hierarchical structures prevailing in social organizations. We determine the mean, variance and kurtosis of the Gaussian-Chain distribution to show its heavy-tailed property, and compute the tail distribution tab…
ADRL improves participant selection in MCS systems.
problem Designing a participant selection algorithm for different MCS systems with multiple goals.
method Auxiliary-task based deep reinforcement learning (ADRL) using transformers and pointer networks.
result ADRL outperforms other baselines in various MCS settings.
Paper introduces IGMM-GAN for multimodal anomaly detection in mobility data.
problem Lack of ground truth data and dependence on pre-processing for anomaly detection in human mobility.
method Coupled IGMM-GAN for generating realistic synthetic datasets and multimodal anomaly detection.
result IGMM-GAN improves anomaly detection performance over existing GAN methods.
GCF estimates heterogeneous treatment effects for continuous treatments in online marketplaces.
problem Estimating heterogeneous treatment effects for continuous treatments in online marketplaces.
method Kernel-based doubly robust estimator and distance-based splitting criterion.
result GCF estimates heterogeneous treatment effects for continuous treatments effectively.
In Part III of this study, we apply the price dynamical model with big buyers and big sellers developed in Part I of this paper to the daily closing prices of the top 20 banking and real estate stocks listed in the Hong Kong Stock Exchange. The basic idea is to estimate the strength parameters of the big buyers and the…
Paper evaluates dynamic QTE for ridesharing data.
problem Assessing QTE in ridesharing with skewed outcomes.
method Developed VCDP models to estimate dynamic CQTE.
result Dynamic CQTE equals sum of individual CQTEs.
Paper predicts urban dispersal events using deep survival analysis on mobility data.
problem Predicting abnormal dispersal events in urban areas to mitigate congestion and safety risks.
method Formulated as a survival analysis problem, developed a two-stage deep learning framework (DILSA).
result DILSA predicts dispersal events with F1-score of 0.7 and average time error of 18 minutes.
Model optimal liquidation in asset bubbles with varying entry times.
problem Optimal liquidation in asset bubbles with variable entry times and exogenous crashes.
method Mean field game (MFG) with varying entry times and progressive enlargement of filtrations.
result Existence of MFG equilibria and decomposition of equilibrium strategies.
New framework validates counterfactual estimations in network interference settings.
problem Challenges in causal effect estimation and validation in network interference settings.
method Introduces a distribution-preserving network bootstrap and counterfactual cross-validation procedure.
result Validates counterfactual estimations in diverse network interference settings.
This study predicts parking availability using multi-source data and a self-supervised learning enhanced transformer.
problem Accurate parking availability prediction to support urban planning and management.
method Proposes SST-iTransformer, a self-supervised learning enhanced spatio-temporal inverted transformer, integrating multi-source data.
result SST-iTransformer achieves state-of-the-art performance in parking availability prediction.
Paper uses machine learning to model travel mode switching under a new transit system.
problem Modeling individual travel mode preferences and response to new mobility options.
method Interpretable machine learning approach to predict and interpret mode-switching behavior.
result Machine learning captures individual heterogeneity in travel mode choice.
Synthetic control method improves policy evaluation in high-dimensional settings.
problem Evaluating the impact of new policies in large-scale applications.
method Two-phase approach: nearest neighbor matching followed by supervised learning.
result The method successfully improves estimate accuracy in large-scale experiments.
In this short note, using our geometric method introduced in a previous paper \cite{phl} and initiated by \cite{ave}, we derive an asymptotic swaption implied volatility at the first-order for a general stochastic volatility Libor Market Model. This formula is useful to quickly calibrate a model to a full swaption matr…
Forecaster uses graph Transformers to forecast spatial and time-dependent data.
problem Complex spatial and temporal dependencies in data.
method Graph Transformer architecture with sparsification for spatial and temporal dependencies.
result Forecaster significantly outperforms state-of-the-art baselines in taxi demand forecasting.
Investors with high risk aversion always invest during financial bubbles.
problem Optimal investment in a financial bubble model.
method Modeling financial bubbles using strict local martingales and Johansen-Ledoit-Sornette (JLS) model relaxations.
result Investors with high relative risk aversion always invest during financial bubbles.
The paper examines how share buybacks impact a company's earnings per share.
problem The trade-off between reducing share count and decreasing net earnings due to share buybacks.
method Review of accretive share repurchases, analysis of EPS increase as a function of price paid, and quantification of earnings growth difference.
result Share buybacks can enhance EPS, but the net effect on earnings growth is mixed.
A method to identify new classes of price jumps in financial markets.
problem Separating endogenous and exogenous causes of price jumps.
method Wavelet-based representation of jump time-series.
result Identification of new classes of jumps and investigation of co-jumps.
A framework for anonymized risk sharing without revealing identities or preferences.
problem Risk sharing without revealing individual identities or preferences.
method Axiomatic framework with four key axioms: actuarial fairness, risk fairness, risk anonymity, and operational anonymity.
result The conditional mean risk sharing rule is uniquely characterized by these axioms.
Study on optimizing task allocation for agents receiving proposals sequentially.
problem Optimizing task allocation for agents receiving proposals sequentially.
method An agent receives task proposals sequentially and can either accept or reject a proposal. The study considers two scenarios: known reward function but unknown task duration distribution, and unknown reward function.
result Regret incurred by the agent in both scenarios.
Defends classifiers from adversarial attacks using self-supervised data estimation.
problem Protecting classifiers from adversarial attacks with full attacker access.
method RIDE, a self-supervised learning algorithm for individual data estimation.
result Significant improvement in adversarial defense performance (98%, 76%, 43% test accuracy on MNIST, CIFAR-10, and ImageNet datasets respectively).