Denis Larocque

Tianyu Shi, Francois-Xavier Devailly, Denis Larocque et al.

A number of deep reinforcement-learning (RL) approaches propose to control traffic signals. In this work, we study the robustness of such methods along two axes. First, sensor failures and GPS occlusions create missing-data challenges and we show that recent methods remain brittle in the face of these missing data. Second, we provide a more systematic study of the generalization ability of RL methods to new networks with different traffic regimes. Again, we identify the limitations of recent approaches. We then propose using a combination of distributional and vanilla reinforcement learning through a policy ensemble. Building upon the state-of-the-art previous model which uses a decentralized approach for large-scale traffic signal control with graph convolutional networks (GCNs), we first learn models using a distributional reinforcement learning (DisRL) approach. In particular, we use implicit quantile networks (IQN) to model the state-action return distribution with quantile regression. For traffic signal control problems, an ensemble of standard RL and DisRL yields superior performance across different scenarios, including different levels of missing sensor data and traffic flow patterns. Furthermore, the learning scheme of the resulting model can improve zero-shot transferability to different road network structures, including both synthetic networks and real-world networks (e.g., Luxembourg, Manhattan). We conduct extensive experiments to compare our approach to multi-agent reinforcement learning and traditional transportation approaches. Results show that the proposed method improves robustness and generalizability in the face of missing data, varying road networks, and traffic flows.

François-Xavier Devailly, Denis Larocque, Laurent Charlin

Most reinforcement learning methods for adaptive-traffic-signal-control require training from scratch to be applied on any new intersection or after any modification to the road network, traffic distribution, or behavioral constraints experienced during training. Considering 1) the massive amount of experience required to train such methods, and 2) that experience must be gathered by interacting in an exploratory fashion with real road-network-users, such a lack of transferability limits experimentation and applicability. Recent approaches enable learning policies that generalize for unseen road-network topologies and traffic distributions, partially tackling this challenge. However, the literature remains divided between the learning of cyclic (the evolution of connectivity at an intersection must respect a cycle) and acyclic (less constrained) policies, and these transferable methods 1) are only compatible with cyclic constraints and 2) do not enable coordination. We introduce a new model-based method, MuJAM, which, on top of enabling explicit coordination at scale for the first time, pushes generalization further by allowing a generalization to the controllers' constraints. In a zero-shot transfer setting involving both road networks and traffic settings never experienced during training, and in a larger transfer experiment involving the control of 3,971 traffic signal controllers in Manhattan, we show that MuJAM, using both cyclic and acyclic constraints, outperforms domain-specific baselines as well as another transferable approach.

Cansu Alakus, Denis Larocque, Aurelie Labbe

Like many predictive models, random forests provide point predictions for new observations. Besides the point prediction, it is important to quantify the uncertainty in the prediction. Prediction intervals provide information about the reliability of the point predictions. We have developed a comprehensive R package, RFpredInterval, that integrates 16 methods to build prediction intervals with random forests and boosted forests. The set of methods implemented in the package includes a new method to build prediction intervals with boosted forests (PIBF) and 15 method variations to produce prediction intervals with random forests, as proposed by Roy and Larocque (2020). We perform an extensive simulation study and apply real data analyses to compare the performance of the proposed method to ten existing methods for building prediction intervals with random forests. The results show that the proposed method is very competitive and, globally, outperforms competing methods.

Cansu Alakus, Denis Larocque, Sebastien Jacquemont et al.

Investigating the relationships between two sets of variables helps to understand their interactions and can be done with canonical correlation analysis (CCA). However, the correlation between the two sets can sometimes depend on a third set of covariates, often subject-related ones such as age, gender, or other clinical measures. In this case, applying CCA to the whole population is not optimal and methods to estimate conditional CCA, given the covariates, can be useful. We propose a new method called Random Forest with Canonical Correlation Analysis (RFCCA) to estimate the conditional canonical correlations between two sets of variables given subject-related covariates. The individual trees in the forest are built with a splitting rule specifically designed to partition the data to maximize the canonical correlation heterogeneity between child nodes. We also propose a significance test to detect the global effect of the covariates on the relationship between two sets of variables. The performance of the proposed method and the global significance test is evaluated through simulation studies that show it provides accurate canonical correlation estimations and well-controlled Type-1 error. We also show an application of the proposed method with EEG data.

Weichi Yao, Halina Frydman, Denis Larocque et al.

Survival data with time-varying covariates are common in practice. If relevant, they can improve on the estimation of survival function. However, the traditional survival forests - conditional inference forest, relative risk forest and random survival forest - have accommodated only time-invariant covariates. We generalize the conditional inference and relative risk forests to allow time-varying covariates. We also propose a general framework for estimation of a survival function in the presence of time-varying covariates. We compare their performance with that of the Cox model and transformation forest, adapted here to accommodate time-varying covariates, through a comprehensive simulation study in which the Kaplan-Meier estimate serves as a benchmark, and performance is compared using the integrated L2 difference between the true and estimated survival functions. In general, the performance of the two proposed forests substantially improves over the Kaplan-Meier estimate. Taking into account all other factors, under the proportional hazard (PH) setting, the best method is always one of the two proposed forests, while under the non-PH setting, it is the adapted transformation forest. K-fold cross-validation is used as an effective tool to choose between the methods in practice.

François-Xavier Devailly, Denis Larocque, Laurent Charlin

Scaling adaptive traffic-signal control involves dealing with combinatorial state and action spaces. Multi-agent reinforcement learning attempts to address this challenge by distributing control to specialized agents. However, specialization hinders generalization and transferability, and the computational graphs underlying neural-networks architectures -- dominating in the multi-agent setting -- do not offer the flexibility to handle an arbitrary number of entities which changes both between road networks, and over time as vehicles traverse the network. We introduce Inductive Graph Reinforcement Learning (IG-RL) based on graph-convolutional networks which adapts to the structure of any road network, to learn detailed representations of traffic-controllers and their surroundings. Our decentralized approach enables learning of a transferable-adaptive-traffic-signal-control policy. After being trained on an arbitrary set of road networks, our model can generalize to new road networks, traffic distributions, and traffic regimes, with no additional training and a constant number of parameters, enabling greater scalability compared to prior methods. Furthermore, our approach can exploit the granularity of available data by capturing the (dynamic) demand at both the lane and the vehicle levels. The proposed method is tested on both road networks and traffic settings never experienced during training. We compare IG-RL to multi-agent reinforcement learning and domain-specific baselines. In both synthetic road networks and in a larger experiment involving the control of the 3,971 traffic signals of Manhattan, we show that different instantiations of IG-RL outperform baselines.