Search

Your search keyword '"Qin, Zhizhen"' showing total 30 results
Start Over Author "Qin, Zhizhen"
30 results on '"Qin, Zhizhen"'

1. SEEV: Synthesis with Efficient Exact Verification for ReLU Neural Barrier Functions

Author

Zhang, Hongchao, Qin, Zhizhen, Gao, Sicun, and Clark, Andrew

Subjects
Electrical Engineering and Systems Science - Systems and Control, Computer Science - Robotics
Abstract

Neural Control Barrier Functions (NCBFs) have shown significant promise in enforcing safety constraints on nonlinear autonomous systems. State-of-the-art exact approaches to verifying safety of NCBF-based controllers exploit the piecewise-linear structure of ReLU neural networks, however, such approaches still rely on enumerating all of the activation regions of the network near the safety boundary, thus incurring high computation cost. In this paper, we propose a framework for Synthesis with Efficient Exact Verification (SEEV). Our framework consists of two components, namely (i) an NCBF synthesis algorithm that introduces a novel regularizer to reduce the number of activation regions at the safety boundary, and (ii) a verification algorithm that exploits tight over-approximations of the safety conditions to reduce the cost of verifying each piecewise-linear segment. Our simulations show that SEEV significantly improves verification efficiency while maintaining the CBF quality across various benchmark systems and neural network structures. Our code is available at https://github.com/HongchaoZhang-HZ/SEEV.

Published
2024

2. Sample-and-Bound for Non-Convex Optimization

Author

Zhai, Yaoguang, Qin, Zhizhen, and Gao, Sicun

Subjects
Computer Science - Artificial Intelligence
Abstract

Standard approaches for global optimization of non-convex functions, such as branch-and-bound, maintain partition trees to systematically prune the domain. The tree size grows exponentially in the number of dimensions. We propose new sampling-based methods for non-convex optimization that adapts Monte Carlo Tree Search (MCTS) to improve efficiency. Instead of the standard use of visitation count in Upper Confidence Bounds, we utilize numerical overapproximations of the objective as an uncertainty metric, and also take into account of sampled estimates of first-order and second-order information. The Monte Carlo tree in our approach avoids the usual fixed combinatorial patterns in growing the tree, and aggressively zooms into the promising regions, while still balancing exploration and exploitation. We evaluate the proposed algorithms on high-dimensional non-convex optimization benchmarks against competitive baselines and analyze the effects of the hyper parameters., Comment: Published at AAAI 2024. Code is available at https://github.com/aaucsd/MCIR

Published
2024

3. Development of a SciFi-based beam monitor for COMET

Author

Xu, Yu, Ning, Yunsong, Qin, Zhizhen, Teng, Yao, Feng, Changqing, Tang, Jian, Chen, Yu, Fukao, Yoshinori, Mihara, Satoshi, and Oishi, Kou

Subjects
Physics - Instrumentation and Detectors, High Energy Physics - Experiment
Abstract

COMET is a leading experiment to search for coherent conversion of $\mu^- \mathrm{N}\to e^- \mathrm{N}$ with a high-intensity pulsed muon beamline, produced by the innovative slow extraction techniques. Therefore, it is critical to measure the characteristics of the muon beam. We set up a Muon Beam Monitor (MBM), where scintillation fibers (SciFi) weaved in the cross shape are coupled to silicon photomultipliers (SiPM), to measure the spatial profile and timing structure of the extracted muon beam for COMET. The MBM detector has been tested successfully with a proton beamline in China Spallation Neutron Source (CSNS) and taken data with good performance in the commissioning run called COMET Phase-$\alpha$. Experience of the MBM development, such as the mechanical structure and electronics readout, and its beam measurement results will be shared.

Published
2023

4. Patching Approximately Safe Value Functions Leveraging Local Hamilton-Jacobi Reachability Analysis

Author

Tonkens, Sander, Toofanian, Alex, Qin, Zhizhen, Gao, Sicun, and Herbert, Sylvia

Subjects
Computer Science - Robotics, Electrical Engineering and Systems Science - Systems and Control
Abstract

Safe value functions, such as control barrier functions, characterize a safe set and synthesize a safety filter, overriding unsafe actions, for a dynamic system. While function approximators like neural networks can synthesize approximately safe value functions, they typically lack formal guarantees. In this paper, we propose a local dynamic programming-based approach to "patch" approximately safe value functions to obtain a safe value function. This algorithm, HJ-Patch, produces a novel value function that provides formal safety guarantees, yet retains the global structure of the initial value function. HJ-Patch modifies an approximately safe value function at states that are both (i) near the safety boundary and (ii) may violate safety. We iteratively update both this set of "active" states and the value function until convergence. This approach bridges the gap between value function approximation methods and formal safety through Hamilton-Jacobi (HJ) reachability, offering a framework for integrating various safety methods. We provide simulation results on analytic and learned examples, demonstrating HJ-Patch reduces the computational complexity by 2 orders of magnitude with respect to standard HJ reachability. Additionally, we demonstrate the perils of using approximately safe value functions directly and showcase improved safety using HJ-Patch., Comment: 8 pages, IEEE Conference on Decision and Control (CDC), 2024 (In Press)

Published
2023

5. Policy Optimization with Advantage Regularization for Long-Term Fairness in Decision Systems

Author

Yu, Eric Yang, Qin, Zhizhen, Lee, Min Kyung, and Gao, Sicun

Subjects
Computer Science - Machine Learning, Computer Science - Artificial Intelligence, Computer Science - Computers and Society
Abstract

Long-term fairness is an important factor of consideration in designing and deploying learning-based decision systems in high-stake decision-making contexts. Recent work has proposed the use of Markov Decision Processes (MDPs) to formulate decision-making with long-term fairness requirements in dynamically changing environments, and demonstrated major challenges in directly deploying heuristic and rule-based policies that worked well in static environments. We show that policy optimization methods from deep reinforcement learning can be used to find strictly better decision policies that can often achieve both higher overall utility and less violation of the fairness requirements, compared to previously-known strategies. In particular, we propose new methods for imposing fairness requirements in policy optimization by regularizing the advantage evaluation of different actions. Our proposed methods make it easy to impose fairness constraints without reward engineering or sacrificing training efficiency. We perform detailed analyses in three established case studies, including attention allocation in incident monitoring, bank loan approval, and vaccine distribution in population networks., Comment: 14 pages, published in NeurIPS 2022

Published
2022

6. Quantifying Safety of Learning-based Self-Driving Control Using Almost-Barrier Functions

Author

Qin, Zhizhen, Weng, Tsui-Wei, and Gao, Sicun

Subjects
Computer Science - Robotics
Abstract

Path-tracking control of self-driving vehicles can benefit from deep learning for tackling longstanding challenges such as nonlinearity and uncertainty. However, deep neural controllers lack safety guarantees, restricting their practical use. We propose a new approach of learning almost-barrier functions, which approximately characterizes the forward invariant set for the system under neural controllers, to quantitatively analyze the safety of deep neural controllers for path-tracking. We design sampling-based learning procedures for constructing candidate neural barrier functions, and certification procedures that utilize robustness analysis for neural networks to identify regions where the barrier conditions are fully satisfied. We use an adversarial training loop between learning and certification to optimize the almost-barrier functions. The learned barrier can also be used to construct online safety monitors through reachability analysis. We demonstrate effectiveness of our methods in quantifying safety of neural controllers in various simulation environments, ranging from simple kinematic models to the TORCS simulator with high-fidelity vehicle dynamics simulation.

Published
2022

9. Performance Analysis of Multi-Tote Storage and Retrieval Autonomous Mobile Robot Systems.

Author

Qin, Zhizhen, Yang, Peng, Gong, Yeming, and de Koster, René B. M.

Subjects
*AUTONOMOUS robots, *MOBILE robots, *GRAPH algorithms, *DYNAMIC programming, *SHELVING (Furniture)
Abstract

Multi-tote storage and retrieval (MTSR) autonomous mobile robots can carry multiple product totes, store and retrieve them from different shelf rack tiers, and transport them to a workstation where the products are picked to fulfill customer orders. In each robot trip, totes retrieved during the previous trip must be stored. This leads to a mixed storage and retrieval route. We analyze this mixed storage and retrieval route problem and derive the optimal travel route for a multiblock warehouse by a layered graph algorithm, based on storage first-retrieval second and mixed storage and retrieval policies. We also propose an effective heuristic routing policy, the closest retrieval (CR) sequence policy, based on a local shortest path. Numerical results show that the CR policy leads to shorter travel times than the well-known S-shape policy, whereas the gap with the optimal mixed storage and retrieval policy in practical scenarios is small. Based on the CR policy, we model the stochastic behavior of the system using a semiopen queuing network (SOQN). This model can accurately estimate average tote throughput time and system throughput capacity as a function of the number of robots in the system. We use the SOQN and corresponding closed queuing network models to optimize the total annual cost as a function of the warehouse shape, the number of robots, and tote buffer positions on the robots for a given average tote throughput time and throughput capacity. Compared with robots that retrieve a single tote per trip, an MTSR system with at least five buffer positions can achieve lower operational costs while meeting given average tote throughput time and tote throughput capacity constraints. Funding: This work was supported by National Natural Science Foundation of China [Grant 72372088] and the Shenzhen Science and Technology Program [Grant GJHZ20220913143003006]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2023.0397. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

11. Impaired Thyroid Hormone Sensitivity Increases Risk of Cardiovascular Events in Patients Undergoing Coronary Angiography.

Author

Qin, Zhizhen, Muhanhali, Dilidaer, and Ling, Yan

Subjects
MYOCARDIAL infarction, CORONARY angiography, THYROID hormones, CARDIOVASCULAR diseases risk factors, THYROID hormone receptors, MAJOR adverse cardiovascular events, PROPORTIONAL hazards models
Abstract

Context Previous studies found inconsistent results on the relationship between thyroid function and cardiovascular risks. Objective This study aimed to investigate the association of thyroid hormone sensitivity with the risk of major adverse cardiovascular events (MACE) and cardiovascular death in a euthyroid population undergoing coronary angiography. Methods This prospective cohort study enrolled 1470 euthyroid participants who underwent coronary angiography between March and November of 2013. The participants were followed up from July to November in 2022. Thyrotroph thyroxine resistance index (TT4RI), TSH index (TSHI), and feedback quantile-based index (TFQI) were calculated to evaluate the sensitivity to thyroid hormone. Kaplan–Meier curve and multivariable Cox proportional hazard model were performed to analyze the association between thyroid hormone sensitivity and risk of MACE and cardiovascular death. Results Among 1089 participants who completed the follow-up, 342 cases of MACE and 77 cardiovascular deaths were identified during a medium follow-up duration of 111 months. In the multivariable Cox proportional hazard model, the higher levels of TFQI (hazard ratio [HR] = 1.41; 95% CI, 1.08-1.84; P for trend =.01), TT4RI (HR = 1.40; 95% CI, 1.06-1.84; P for trend =.02) and TSHI (HR = 1.61; 95% CI, 1.22-2.13; P for trend =.001) were associated with increased risk of MACE. The higher levels of TFQI (HR = 2.21; 95% CI, 1.17-4.17; P for trend =.02) and TSHI (HR = 2.05; 95% CI; 1.08-3.91; P for trend =.03) were also associated with increased risk of cardiovascular death. Conclusion Impaired sensitivity to thyroid hormone is associated with higher risks of MACE and cardiovascular death in a euthyroid population undergoing coronary angiography. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

14. Learning Safety Certificates for Neural Controllers

Author

Qin, Zhizhen

Subjects
Artificial intelligence, Robotics, Computer science, Deep Learning, Robotics, Safety
Abstract

Learning-based methods are promising for tackling the inherent nonlinearity and model uncertainty in path-tracking control problems, but the lack of safety guarantee of neural controllers restricts their practical use in self-driving vehicles. We propose methods for training and certifying barrier functions that are themselves represented as neural networks, for ensuring safety properties of learning-based neural controllers in self-driving with realistic nonlinear dynamics. We describe how to identify safe and unsafe regions of the state space and minimize the violation of the barrier conditions to train neural barrier functions. We then show how to leverage the recent advances in robustness analysis of neural networks to bound the Lie derivatives of the barrier functions and certify the barrier conditions. Although understanding the vehicle dynamics is important for designing the training and certification procedures, the proposed algorithms are sampling-based and do not assume access to analytic forms of the dynamics. Thus the methods are generally applicable to nonlinear vehicle dynamics with model uncertainty and partial observability. We demonstrate the effectiveness of the methods for quantitatively certifying safety of neural controllers in different simulation environments ranging from simple kinematic models to the TORCS high-fidelity vehicle dynamics modeling environment as a blackbox simulator.

Published
2021

15. Dual-Salts Electrolyte with Fluoroethylene Carbonate Additive for High-Voltage Li-Metal Batteries

Author

Qin, Zhizhen, Wu, Baolin, Danilov, Dmitri L., Eichel, Rüdiger A., Notten, Peter H.L., Qin, Zhizhen, Wu, Baolin, Danilov, Dmitri L., Eichel, Rüdiger A., and Notten, Peter H.L.

Abstract

The combination of Li-metal anode and high-voltage cathode is regarded as a solution for the next-generation high-energy-density secondary batteries. However, a traditional electrolyte is either incompatible with the Li-metal anode or vulnerable to high voltage. This work reports a 1 M dual-salts Localized-High-Concentration-Electrolyte with Fluoroethylene carbonate (FEC) additive. It enables stable cycling of Li||LiNi0.8Co0.1Mn0.1O2 (NMC811) battery, which shows 81.5% capacity retention after 300 cycles with a charge/discharge current density of 1 C and a voltage range of 2.7–4.4 V. Scanning electron microscopy (SEM) images show that this electrolyte not only largely reduced Li dendrites and ‘dead’ Li on anode surface but also well protected the microstructure of NMC811 cathode. Possible components of both solid-electrolyte interlayer (SEI) and cathode-electrolyte interlayer (CEI) were characterized by energy-dispersive X-ray spectroscopy (EDX). The result illustrates that FEC protected Li salts from decomposition on the anode side and suppressed the decomposition of solvents on the cathode side.

Published
2023

16. Patching Neural Barrier Functions Using Hamilton-Jacobi Reachability

Author

Tonkens, Sander, Toofanian, Alex, Qin, Zhizhen, Gao, Sicun, Herbert, Sylvia, Tonkens, Sander, Toofanian, Alex, Qin, Zhizhen, Gao, Sicun, and Herbert, Sylvia

Abstract

Learning-based control algorithms have led to major advances in robotics at the cost of decreased safety guarantees. Recently, neural networks have also been used to characterize safety through the use of barrier functions for complex nonlinear systems. Learned barrier functions approximately encode and enforce a desired safety constraint through a value function, but do not provide any formal guarantees. In this paper, we propose a local dynamic programming (DP) based approach to "patch" an almost-safe learned barrier at potentially unsafe points in the state space. This algorithm, HJ-Patch, obtains a novel barrier that provides formal safety guarantees, yet retains the global structure of the learned barrier. Our local DP based reachability algorithm, HJ-Patch, updates the barrier function "minimally" at points that both (a) neighbor the barrier safety boundary and (b) do not satisfy the safety condition. We view this as a key step to bridging the gap between learning-based barrier functions and Hamilton-Jacobi reachability analysis, providing a framework for further integration of these approaches. We demonstrate that for well-trained barriers we reduce the computational load by 2 orders of magnitude with respect to standard DP-based reachability, and demonstrate scalability to a 6-dimensional system, which is at the limit of standard DP-based reachability., Comment: 8 pages, submitted to IEEE Conference on Decision and Control (CDC), 2023

Published
2023

17. Engineering extracellular vesicles with macrophage membranes fusion ameliorated psoriatic skin inflammation in imiquimod‐treated mice

Author

Wang, Zeng, primary, Qin, Zhizhen, additional, Wang, Jiadie, additional, Xu, Xinqi, additional, Zhang, Mengxin, additional, Liang, Yuyue, additional, Huang, Yukun, additional, Yu, Zengyang, additional, Gong, Yu, additional, Zhou, Luxian, additional, Qiu, Yiran, additional, Ma, Minglu, additional, Li, Dan, additional, and Li, Bin, additional

Published
2022
Full Text
View/download PDF

20. Engineering extracellular vesicles with macrophage membrane fusion for ameliorating imiquimod-induced psoriatic skin inflammation.

Author

Wang, Zeng, Qin, Zhizhen, Wang, Jiadie, Xu, Xinqi, Zhang, Mengxin, Liang, Yuyue, Huang, Yukun, Yu, Zengyang, Gong, Yu, Zhou, Luxian, Qiu, Yiran, Ma, Minglu, Li, Dan, and Li, Bin

Subjects
*SKIN inflammation, *EXTRACELLULAR vesicles, *MEMBRANE fusion, *MACROPHAGES, *SUBCUTANEOUS injections
Abstract

Herein, we developed an engineered extracellular vehicle (EV)-based method for ameliorating inflammatory responses in psoriasis. EVs, derived from annexin A1 (ANXA1) overexpressing T cells, were co-extruded with M2 macrophage membrane to obtain engineered EVs. In vitro, the effect of engineered EVs on macrophage polarization was evaluated by real-time PCR. In imiquimod (IMQ)-induced psoriasis-like mouse model, the efficacy of engineered EVs in ameliorating psoriatic inflammation was evaluated by Psoriasis Area and Severity Index (PASI) score and immunohistochemistry staining after subcutaneous injection of EVs. The engineered EVs not only preserved the high stability of M2 macrophage membrane but also retained the macrophage reprogramming potential of ANXA1 overexpressed in T cells. In the psoriasis-like mouse model, subcutaneous injection of engineered EVs successfully reduced the PASI score and the levels of pro-inflammatory cytokines, including IL-1β, IL-6, and TNF-α. Along with high biosafety, the administration of EVs also rescued the histomorphological changes of spleen, liver, and kidney. The engineered EVs exhibited the potential to alleviate inflammation of psoriasis, providing new insights and potential strategies for the immunotherapies of psoriasis. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

30. [Comparison of extent of postoperative hydrocephalus in patients between intervertional therapy with embolism and craniotomy occlusion in Hunt-Hess III-IV level aneurysm induced subarachnoid hemorrhage and their prognosis].

Author

Liu Y, Sun S, Chen X, Cheng S, Qin Z, Liu X, Chen X, Ning L, and Wang Z

Subjects
Aneurysm, Drainage, Humans, Postoperative Period, Prognosis, Retrospective Studies, Craniotomy, Embolization, Therapeutic, Hydrocephalus, Subarachnoid Hemorrhage therapy
Abstract

Objective: To analyze and compare the difference and prognosis between vascular embolization and craniotomy occlusion in patients suffering from aneurysmal subarachnoid hemorrhage (aSAH) with Hunt-Hess level III-IV, and acute postoperative hydrocephalus., Methods: A retrospective study was conducted on 767 patients who had undergone vascular embolization (vascular embolization group, n = 403) or craniotomy occlusion operation (craniotomy occlusion operation group, n = 364), and the patients with postoperative acute hydrocephalus were screened. The clinical data of patients of both groups was analyzed. By judging short-term prognosis in patients with hydrocephalus with Glasgow outcome scale (GOS) score estimated at discharge, the advantages and disadvantages of two surgical procedures were compared., Results: The number of cases with postoperative hydrocephalus in vascular embolization group was 56 (13.90%), while that in craniotomy occlusion group was 33 (9.07%). The difference between the two groups of incidence of hydrocephalus was statistically significant (χ (2) = 4.350, P = 0.037). In 767 patients with aSAH, the incidence of hydrocephalus among the patients after the hematoma removal operation was significantly lower than that of patients without hematoma removal [3.07% (11/358) vs. 19.07% (78/409), χ (2) = 47.635, P = 0.000]. The incidence of hydrocephalus among the patients after ventricular drainage was significantly lower than that of patients without the drainage [2.77% (19/685) vs. 85.37% (70/82), χ (2) = 487.032, P = 0.000]. In 403 cases of vascular embolization group, the incidence of hydrocephalus in the patients after the hematoma removal operation was lower than that of patients without it [8.06% (5/62) vs. 14.96% (51/341), χ (2) = 2.082, P = 0.168]. The incidence of hydrocephalus in the patients after the ventricular drainage was lower than that of patients without drainage [2.59% (9/347) vs. 83.93% (47/56), χ (2) = 266.599, P = 0.000]. In 364 cases of craniotomy occlusion operation group, the incidence of hydrocephalus in the patients after hematoma removal operation was significantly lower than that of patients did not receive [2.03% (6/296) vs. 39.71% (27/68), χ (2) = 95.226, P = 0.000]. The incidence of hydrocephalus among the patients after the ventricular drainage was significantly lower than that of patients without drainage [2.96% (10/338) vs. 88.46% (23/26), χ (2) = 203.852, P = 0.000]. The difference in incidence of hydrocephalus between the patients who had hematoma removal surgery between vascular embolization group and craniotomy occlusion operation group was statistically significant [8.06% (5/62) vs. 2.03% (6/296), χ (2) = 4.411, P = 0.027], while no statistically difference was present in ventricular drainage patients [2.59% (9/347) vs. 2.96% (10/338), χ (2) = 0.085, P = 0.819]. There were 23 patients (41.07%) with good outcome (GOS score 4-5), while 33 (58.93%) with poor outcome (GOS score 1-3) in 56 patients undergone vascular embolization operation. Good result (GOS score 4-5) was shown in 21 (63.64%) and 12 (36.36%) with poor outcome (GOS score 1-3) among 33 patients with hydrocephalus after craniotomy occlusion operation, and the difference was statistically significant (χ (2) = 4.230, P = 0.039)., Conclusions: Hematoma is one of the main factor contributing to the differences in the incidence of postoperative hydrocephalus of Hunt-Hess grade III-IV patients either receiving vascular embolization or craniotomy occlusion operation. Lateral ventricle drainage may not be the factor that contributes to the difference in incidence of hydrocephalus formation between the vascular embolization and craniotomy occlusion operation groups in Hunt-Hess level III-IV patients. The short term prognosis in the craniotomy occlusion operation group is superior to that of endovascular intervention embolization group.

Published
2015
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results