首页/论文/2001.11062

arXiv 2001.110622020-01-29cs.LGstat.ML

Safe Predictors for Enforcing Input-Output Specifications

MIT Lincoln Laboratory（麻省理工学院林肯实验室）

Stephen Mell, Olivia Brown, Justin Goodwin, Sung-Hyun Son

备注：: 10 pages, 5 figures, paper accepted to the NeurIPS 2019 Workshop on Machine Learning with Guarantees and the NeurIPS 2019 Workshop on Safety and Robustness in Decision Making

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

英文摘要相关论文

英文摘要：

We present an approach for designing correct-by-construction neural networks (and other machine learning models) that are guaranteed to be consistent with a collection of input-output specifications before, during, and after algorithm training. Our method involves designing a constrained predictor for each set of compatible constraints, and combining them safely via a convex combination of their predictions. We demonstrate our approach on synthetic datasets and an aircraft collision avoidance problem.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

继续探索

相关论文

Gap Safe screening rules for sparsity enforcing penalties

In high dimensional regression settings, sparsity enforcing penalties have proved useful to regularize the data-fitting term. A recently introduced technique called screening rules propose to ignore some variables in the optimization leveraging the expected sparsity of the solutions and consequently leading to faster solvers. When the procedure is guaranteed not to discard variables wrongly the rules are said to be safe. In this work, we propose a unifying framework for generalized linear models regularized with standard sparsity enforcing penalties such as $\ell_1$ or $\ell_1/\ell_2$ norms. O…

2017-12-27cs.LGmath.OCstat.CO

用软屏障强制执行硬约束：未知随机环境中的安全强化学习

Enforcing Hard Constraints with Soft Barriers: Safe Reinforcement Learning in Unknown Stochastic Environments

Northwestern University（西北大学）；University of California, Berkeley（加州大学伯克利分校）；University of Pennsylvania（宾夕法尼亚大学）；Yale University（耶鲁大学）；University of Liverpool（利物浦大学）

针对未知随机环境下强化学习智能体的硬安全约束难满足的问题，提出基于生成模型软屏障的安全RL方法，联合学习环境与优化策略，仿真安全率显著优于CMDP基线方法。

2023-06-13cs.LGcs.SYeess.SY

在深度学习模型中强制执行硬线性约束的决策规则

Enforcing Hard Linear Constraints in Deep Learning Models with Decision Rules

Davidson School of Chemical Engineering（戴维森化学工程学院）；Purdue University（普渡大学）

本文提出了一种模型无关的框架，通过结合任务网络和安全网络，强制深度学习模型在训练和推理过程中满足输入依赖的线性约束，同时保持高准确性和低计算成本。

2025-11-24cs.LG

SAFE-SVD：面向物理基础模型的敏感性感知保真度压缩SVD

SAFE-SVD: Sensitivity-Aware Fidelity-Enforcing SVD for Physics Foundation Models

AI Centre, University College London（伦敦大学学院人工智能中心）；University College London（伦敦大学学院）；Central South University（中南大学）；University of Massachusetts at Amherst（马萨诸塞大学阿姆赫斯特分校）

本文提出了一种新的压缩物理基础模型的方法，通过在压缩过程中显式建模损失感知的层敏感性，以保持准确性和物理保真度，实验表明在多个模型和数据集上实现了显著的压缩增益。

2026-08-14cs.AIcs.LG

Verifiably Safe Exploration for End-to-End Reinforcement Learning

Massachusetts Institute of Technology（麻省理工学院）；IBM Research（IBM研究院）

Deploying deep reinforcement learning in safety-critical settings requires developing algorithms that obey hard constraints during exploration. This paper contributes a first approach toward enforcing formal safety constraints on end-to-end policies with visual inputs. Our approach draws on recent advances in object detection and automated reasoning for hybrid dynamical systems. The approach is evaluated on a novel benchmark that emphasizes the challenge of safely exploring in the presence of hard constraints. Our benchmark draws from several proposed problem sets for safe learning and include…

2020-07-02cs.AIcs.LGcs.LO

基于技能的安全强化学习与风险规划

Skill-based Safe Reinforcement Learning with Risk Planning

School of Computer Science, Carleton University（卡尔顿大学计算机科学学院）

本文提出安全技能规划（SSkP）方法，利用离线演示数据通过PU学习构建技能风险预测器，并结合在线风险规划高效学习风险规避的安全策略，在多个机器人仿真基准中优于现有安全强化学习方法。

2025-05-02cs.AIcs.LG

更多相关论文收起更多论文

在多输出概率回归中通过预秩正则化强制校准

Enforcing Calibration in Multi-Output Probabilistic Regression with Pre-rank Regularization

本研究提出一种通用正则化框架，通过惩罚预秩概率积分变换偏离均匀分布，在训练中强制多输出概率回归的多元校准，并引入PCA预秩，在18个数据集上显著提升校准且不损精度。

2025-10-27cs.LGstat.ML

Predicting Model Failure using Saliency Maps in Autonomous Driving Systems

Texas A&M University（得克萨斯农工大学）

While machine learning systems show high success rate in many complex tasks, research shows they can also fail in very unexpected situations. Rise of machine learning products in safety-critical industries cause an increase in attention in evaluating model robustness and estimating failure probability in machine learning systems. In this work, we propose a design to train a student model -- a failure predictor -- to predict the main model's error for input instances based on their saliency map. We implement and review the preliminary results of our failure predictor model on an autonomous vehi…

2019-05-19cs.LGstat.ML

Look-Ahead Screening Rules for the Lasso（套索的前瞻筛选规则）

Look-Ahead Screening Rules for the Lasso

Lund University（隆德大学）

针对高维场景下套索求解计算成本高的问题，本文提出前瞻筛选策略，利用安全筛选规则确定预测变量无法入模的惩罚值范围以沿解路径筛选变量，实验证明其性能优于Gap Safe规则的主动热启动版本。

2021-06-29cs.LGstat.COstat.ML

CG4AI：一种用于在约束条件下训练AI模型的列生成框架

CG4AI: A Column Generation Framework for Training AI Models Under Constraints

Huawei Technologies Ltd., France Research Center（华为技术有限公司法国研究中心）

CG4AI是带线性约束的AI模型训练的列生成框架，通过主LP、定价子问题和割平面程序实现，在MNIST和SNDLIB网络上验证，可生成可行预测器且准确率优于单模型基线。

2026-08-28cs.AIcs.DMcs.LG

使用 Random Forest 预测器进行美国道路事故预测

United States Road Accident Prediction using Random Forest Predictor

Computer Science and Engineering University at Buffalo（计算机科学与工程大学布法罗分校）

本文基于覆盖美国49州的多源交通数据，结合回归、时间序列分析及 Random Forest 等模型预测事故数量，并通过时空分析识别高风险区域，以支持资源配置和道路安全干预决策。

2025-04-28cs.AIcs.CYcs.LG

Enforcing robust control guarantees within neural network policies

Carnegie Mellon University（卡内基梅隆大学）；Johns Hopkins University（约翰斯·霍普金斯大学）；Bosch Center for AI（博世人工智能中心）

When designing controllers for safety-critical systems, practitioners often face a challenging tradeoff between robustness and performance. While robust control methods provide rigorous guarantees on system stability under certain worst-case disturbances, they often yield simple controllers that perform poorly in the average (non-worst) case. In contrast, nonlinear control methods trained using deep learning have achieved state-of-the-art performance on many control tasks, but often lack robustness guarantees. In this paper, we propose a technique that combines the strengths of these two appro…

2021-01-28cs.LGmath.OC

谱对齐作为神经网络训练中损失爆炸的预测指标

Spectral Alignment as Predictor of Loss Explosion in Neural Network Training

Tsinghua University（清华大学）；Beijing Institute of Mathematical Sciences and Applications（北京数学科学研究院）

针对神经网络训练中损失爆炸难以及时预警的问题，提出具有理论基础的谱对齐（SA）指标，通过监测层输入与权重矩阵主奇异向量的分布对齐符号多样性崩塌，实现比传统指标更早更清晰的损失爆炸预警，且计算开销低，实用性强。

2025-10-05cs.LG

Rare-Event Simulation for Neural Network and Random Forest Predictors

Columbia University（哥伦比亚大学）；Carnegie Mellon University（卡内基梅隆大学）

We study rare-event simulation for a class of problems where the target hitting sets of interest are defined via modern machine learning tools such as neural networks and random forests. This problem is motivated from fast emerging studies on the safety evaluation of intelligent systems, robustness quantification of learning models, and other potential applications to large-scale simulation in which machine learning tools can be used to approximate complex rare-event set boundaries. We investigate an importance sampling scheme that integrates the dominating point machinery in large deviations …

2020-10-10cs.LGmath.STstat.ML

Neural Networks for Safety-Critical Applications - Challenges, Experiments and Perspectives

fortiss - Landesforschungsinstitut des Freistaats Bayern（巴伐利亚州fortiss研究机构）

We propose a methodology for designing dependable Artificial Neural Networks (ANN) by extending the concepts of understandability, correctness, and validity that are crucial ingredients in existing certification standards. We apply the concept in a concrete case study in designing a high-way ANN-based motion predictor to guarantee safety properties such as impossibility for the ego vehicle to suggest moving to the right lane if there exists another vehicle on its right.

2017-09-04cs.LGcs.SE

基于能量神经网络的安全设计学习

Safe-by-Design Learning via Energy-based Neural Networks

The Italian Institute of Artificial Intelligence for Industry（意大利人工智能工业研究所）；University of Colorado Boulder（科罗拉多大学博尔德分校）；TU Delft（代尔夫特理工大学）

本文提出基于能量现代Hopfield网络与端口哈密顿神经ODE的安全设计架构，通过构造障碍函数保证安全性，在多个基准包括12维纳米无人机上实现最先进性能并提升鲁棒性。

2026-10-01cs.AIcs.LGcs.SY

Safe Reinforcement Learning via Shielding

University of Texas at Austin（德克萨斯大学奥斯汀分校）；IAIK, Graz University of Technology（格拉茨技术大学 IAIK）；University of Bremen（不来梅大学）；DFKI GmbH（德国人工智能研究中心）

Reinforcement learning algorithms discover policies that maximize reward, but do not necessarily guarantee safety during learning or execution phases. We introduce a new approach to learn optimal policies while enforcing properties expressed in temporal logic. To this end, given the temporal logic specification that is to be obeyed by the learning system, we propose to synthesize a reactive system called a shield. The shield is introduced in the traditional learning process in two alternative ways, depending on the location at which the shield is implemented. In the first one, the shield acts …

2017-09-03cs.AIcs.LGcs.LO