We propose a learning-based robust predictive control algorithm that compensates for significant uncertainty in the dynamics for a class of discrete-time systems that are nominally linear with an additive nonlinear component. Such systems commonly model the nonlinear effects of an unknown environment on a nominal system. We optimize over a class of nonlinear feedback policies inspired by certainty equivalent "estimate-and-cancel" control laws pioneered in classical adaptive control to achieve significant performance improvements in the presence of uncertainties of large magnitude, a setting in which existing learning-based predictive control algorithms often struggle to guarantee safety. In contrast to previous work in robust adaptive MPC, our approach allows us to take advantage of structure (i.e., the numerical predictions) in the a priori unknown dynamics learned online through function approximation. Our approach also extends typical nonlinear adaptive control methods to systems with state and input constraints even when we cannot directly cancel the additive uncertain function from the dynamics. Moreover, we apply contemporary statistical estimation techniques to certify the system’s safety through persistent constraint satisfaction with high probability. Finally, we show in simulation that our method can accommodate more significant unknown dynamics terms than existing methods.
more »
« less
Parameterizing Lognormal state space models using moment matching
In ecology, it is common for processes to be bounded based on physical constraints of the system. One common example is the positivity constraint, which applies to phenomena such as duration times, population sizes, and total stock of a system’s commodity. In this paper, we propose a novel method for parameterizing Lognormal state space models using an approach based on moment matching. Our method enforces the positivity constraint, allows for arbitrary mean evolution and variance structure, and has a closed-form Markov transition density which allows for more flexibility in fitting techniques. We discuss two existing Lognormal state space models and examine how they differ from the method presented here. We use 180 synthetic datasets to compare the forecasting performance under model misspecification and assess the estimation of precision parameters between our method and existing methods. We find that our models perform well under misspecification, and that fixing the observation variance both helps to improve estimation of the process variance and improves forecast performance. To test our method on a difficult problem, we compare the predictive performance of two Lognormal state space models in predicting the Leaf Area Index over a 151 day horizon by using a process-based ecosystem model to describe the temporal dynamics. We find that our moment matching model performs better than its competitor, and is better suited for intermediate predictive horizons. Overall, our study helps to inform practitioners about the importance of incorporating sensible dynamics when using models of complex systems to predict out-of-sample.
more » « less- NSF-PAR ID:
- 10432070
- Publisher / Repository:
- Springer Science + Business Media
- Date Published:
- Journal Name:
- Environmental and Ecological Statistics
- ISSN:
- 1352-8505
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Current deep neural network approaches for camera pose estimation rely on scene structure for 3D motion estimation, but this decreases the robustness and thereby makes cross-dataset generalization difficult. In contrast, classical approaches to structure from motion estimate 3D motion utilizing optical flow and then compute depth. Their accuracy, however, depends strongly on the quality of the optical flow. To avoid this issue, direct methods have been proposed, which separate 3D motion from depth estimation, but compute 3D motion using only image gradients in the form of normal flow. In this paper, we introduce a network NFlowNet, for normal flow estimation which is used to enforce robust and direct constraints. In particular, normal flow is used to estimate relative camera pose based on the cheirality (depth positivity) constraint. We achieve this by formulating the optimization problem as a differentiable cheirality layer, which allows for end-to-end learning of camera pose. We perform extensive qualitative and quantitative evaluation of the proposed DiffPoseNet’s sensitivity to noise and its generalization across datasets. We compare our approach to existing state-of-the-art methods on KITTI, TartanAir, and TUM-RGBD datasets.more » « less
-
null (Ed.)Background The use of wearables facilitates data collection at a previously unobtainable scale, enabling the construction of complex predictive models with the potential to improve health. However, the highly personal nature of these data requires strong privacy protection against data breaches and the use of data in a way that users do not intend. One method to protect user privacy while taking advantage of sharing data across users is federated learning, a technique that allows a machine learning model to be trained using data from all users while only storing a user’s data on that user’s device. By keeping data on users’ devices, federated learning protects users’ private data from data leaks and breaches on the researcher’s central server and provides users with more control over how and when their data are used. However, there are few rigorous studies on the effectiveness of federated learning in the mobile health (mHealth) domain. Objective We review federated learning and assess whether it can be useful in the mHealth field, especially for addressing common mHealth challenges such as privacy concerns and user heterogeneity. The aims of this study are to describe federated learning in an mHealth context, apply a simulation of federated learning to an mHealth data set, and compare the performance of federated learning with the performance of other predictive models. Methods We applied a simulation of federated learning to predict the affective state of 15 subjects using physiological and motion data collected from a chest-worn device for approximately 36 minutes. We compared the results from this federated model with those from a centralized or server model and with the results from training individual models for each subject. Results In a 3-class classification problem using physiological and motion data to predict whether the subject was undertaking a neutral, amusing, or stressful task, the federated model achieved 92.8% accuracy on average, the server model achieved 93.2% accuracy on average, and the individual model achieved 90.2% accuracy on average. Conclusions Our findings support the potential for using federated learning in mHealth. The results showed that the federated model performed better than a model trained separately on each individual and nearly as well as the server model. As federated learning offers more privacy than a server model, it may be a valuable option for designing sensitive data collection methods.more » « less
-
Nucleic acid kinetic simulators aim to predict the kinetics of interacting nucleic acid strands. Many simulators model the kinetics of interacting nucleic acid strands as continuous-time Markov chains (CTMCs). States of the CTMCs represent a collection of secondary structures, and transitions between the states correspond to the forming or breaking of base pairs and are determined by a nucleic acid kinetic model. The number of states these CTMCs can form may be exponentially large in the length of the strands, making two important tasks challenging, namely, mean first passage time (MFPT) estimation and parameter estimation for kinetic models based on MFPTs. Gillespie’s stochastic simulation algorithm (SSA) is widely used to analyze nucleic acid folding kinetics, but could be computationally expensive for reactions whose CTMC has a large state space or for slow reactions. It could also be expensive for arbitrary parameter sets that occur in parameter estimation. Our work addresses these two challenging tasks, in the full state space of all non-pseudoknotted secondary structures of each reaction. In the first task, we show how to use a reduced variance stochastic simulation algorithm (RVSSA), which is adapted from SSA, to estimate the MFPT of a reaction’s CTMC. In the second task, we estimate model parameters based on MFPTs. To this end, first, we show how to use a generalized method of moments (GMM) approach, where we minimize a squared norm of moment functions that we formulate based on experimental and estimated MFPTs. Second, to speed up parameter estimation, we introduce a fixed path ensemble inference (FPEI) approach, that we adapt from RVSSA. We implement and evaluate RVSSA and FPEI using the Multistrand kinetic simulator. In our experiments on a dataset of DNA reactions, FPEI speeds up parameter estimation compared to inference using SSA, by more than a factor of three for slow reactions. Also, for reactions with large state spaces, it speeds up parameter estimation by more than a factor of two.more » « less
-
more » « less
Summary Covariate-adaptive randomization is popular in clinical trials with sequentially arrived patients for balancing treatment assignments across prognostic factors that may have influence on the response. However, existing theory on tests for the treatment effect under covariate-adaptive randomization is limited to tests under linear or generalized linear models, although the covariate-adaptive randomization method has been used in survival analysis for a long time. Often, practitioners will simply adopt a conventional test to compare two treatments, which is controversial since tests derived under simple randomization may not be valid in terms of type I error under other randomization schemes. We derive the asymptotic distribution of the partial likelihood score function under covariate-adaptive randomization and a working model that is subject to possible model misspecification. Using this general result, we prove that the partial likelihood score test that is robust against model misspecification under simple randomization is no longer robust but conservative under covariate-adaptive randomization. We also show that the unstratified log-rank test is conservative and the stratified log-rank test remains valid under covariate-adaptive randomization. We propose a modification to variance estimation in the partial likelihood score test, which leads to a score test that is valid and robust against arbitrary model misspecification under a large family of covariate-adaptive randomization schemes including simple randomization. Furthermore, we show that the modified partial likelihood score test derived under a correctly specified model is more powerful than log-rank-type tests in terms of Pitman’s asymptotic relative efficiency. Simulation studies about the type I error and power of various tests are presented under several popular randomization schemes.
