We present an algorithmic framework for learning multiple related tasks. Our framework exploits a form of prior knowledge that relates the output spaces of these tasks. We present PAC learning results that analyze the conditions under which such learning is possible. We present results on learning a shallow parser and named-entity recognition system that exploits our framework, showing consistent improvements over baseline methods.
Knowledge Distillation via Route Constrained Optimization
SenseTime Research(商汤科技研究院);National University of Defense Technology(国防科技大学);The Chinese University of Hong Kong(香港中文大学);Beihang University(北京航空航天大学);Tsinghua University(清华大学)
ReFactX: Scalable Reasoning with Reliable Facts via Constrained Generation
University of Milano-Bicocca(米兰-比科卡大学);Banca d'Italia(意大利银行);Institute for Applied Informatics (InfAI)(应用信息研究所);TIB Leibniz Information Centre for Science and Technology(科学与技术信息中心);ScaDS.AI Dresden/Leipzig(ScaDS.AI 德累斯顿/莱比锡分校);Technische Universität Dresden(德累斯顿技术大学);Data Science Institute(数据科学研究所);Leibniz University Hannover(汉诺威莱布尼茨大学);Amazon(亚马逊)
Cross-Learning from Scarce Data via Multi-Task Constrained Optimization
Department of Electrical Engineering, School of Engineering, Universidad de la República(电气工程系,工程学院,乌拉圭共和国大学);Massachusetts Institute of Technology(麻省理工学院);Department of Engineering, Universitat Pompeu Fabra(工程系,庞培法布拉大学);Department of Electrical and Systems Engineering, University of Pennsylvania(电气与系统工程系,宾夕法尼亚大学)
Knowledge Integration in Differentiable Models: A Comparative Study of Data-Driven, Soft-Constrained, and Hard-Constrained Paradigms for Identification and Control of the Single Machine Infinite Bus System
Korea University(高丽大学);Hyundai Motor Company(现代汽车公司)
SC-LoRA: Balancing Efficient Fine-tuning and Knowledge Preservation via Subspace-Constrained LoRA
Shanghai Qi Zhi Institute(上海启智研究院);Institute for Interdisciplinary Information Sciences, Tsinghua University(清华大学交叉信息研究院);Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);Xiongan AI Institute(雄安人工智能研究院)
Near-Optimal Algorithms for Constrained k-Center Clustering with Instance-level Background Knowledge
School of Mathematics and Statistics, Fuzhou University(福州大学数学与统计学学院);School of Accounting, Information Systems and Supply Chain, RMIT University(皇家墨尔本理工大学会计、信息系统与供应链学院);School of Information Technology, Deakin University(德肯大学信息科技学院);Western Sydney University(西悉尼大学);CSIRO’s Data61(澳大利亚联邦科学与工业研究组织Data61)
Why Pretraining Fails to Share Cross-Lingual Knowledge
Weizmann Institute of Science(魏茨曼科学研究所);Bar-Ilan University(巴伊兰大学);Johns Hopkins University(约翰斯·霍普金斯大学);A*STAR(新加坡科技研究局);University of Washington(华盛顿大学);MIT(麻省理工学院);MIT-IBM Watson AI Lab(MIT-IBM沃森人工智能实验室)
Knowledge-Constrained Shape Optimization with a Mixture-of-Experts Neural Operator for High-Confidence Design
School of Ocean and Civil Engineering, Shanghai Jiao Tong University(上海交通大学海洋与土木工程学院);Eastern Institute of Technology(东方理工学院);TenFong Technology Co., Ltd.(十方科技有限公司);IM Motors Technology Co., Ltd.(集度汽车科技有限公司)
Knowledge Distillation and Enhanced Subdomain Adaptation Using Graph Convolutional Network for Resource-Constrained Bearing Fault Diagnosis
Department of Electrical and Computer Engineering, Tarbiat Modares University, Iran(电子与计算机工程系,塔里比亚特·莫达雷斯大学,伊朗);Faculty of Engineering and Technology, University of Mazandaran, Iran(工程与技术学院,马赞德兰大学,伊朗);Department of Medicine, Baylor College of Medicine, USA(医学系,贝勒医学院,美国)
Federated Knowledge Transfer Fine-tuning Large Server Model with Resource-Constrained IoT Clients
School of Intelligent Systems Engineering, Sun Yat-Sen University(中山大学智能工程学院);Shenzhen Fangle Technology Co., Ltd.(深圳方乐科技有限公司);School of Computer Science and Engineering, Nanyang Technological University(南洋理工大学计算机科学与工程学院);School of Computer Science and Technology, Dalian University of Technology(大连理工大学计算机科学与技术学院);School of Electronic Engineering and Computer Science, Queen Mary University of London(伦敦玛丽女王大学电子工程与计算机科学学院)
Rank-Constrained Adaptation for Reliable Real-World Performance
Department of Electrical Engineering \& Computer Science, Massachusetts Institute of Technology, Cambridge, MA, USA;Harvard School of Medicine, Harvard, Boston, MA, USA
HiBid: A Cross-Channel Constrained Bidding System with Budget Allocation by Hierarchical Offline Deep Reinforcement Learning
School of Computer Science and Technology, Beijing Institute of Technology(北京理工大学计算机科学与技术学院);Meituan(美团);Institute for Advanced Algorithms Research(先进算法研究院)
A Locally Penalized Cross-Estimate Federated Method with Guarantees for Constrained Personalized Learning
Department of Industrial and Systems Engineering, Rutgers University(工业与系统工程系,罗杰斯大学);Department of Electrical and Computer Engineering, Rutgers University(电气与计算机工程系,罗杰斯大学)
CramNet: Camera-Radar Fusion with Ray-Constrained Cross-Attention for Robust 3D Object Detection
Waymo(Waymo公司)
稳健的3D目标检测对于安全自动驾驶至关重要。摄像头和雷达传感器是协同作用的,因为它们捕获互补信息,并在不同的环境条件下工作良好。然而,融合相机和雷达数据是具有挑战性的,因为每个传感器都缺乏沿垂直轴的信息,也就是说,相机不知道深度,雷达不知道高度。我们提出了相机-雷达匹配网络CramNet,这是一种有效的方法,可以在联合3D空间中融合相机和雷达的传感器读数。为了更好地利用雷达距离测量相机深度预测,我们提出了一种新的射线约束的交叉注意力机制,解决了相机特征和雷达特征之间的几何对应关系的模糊性。我们的方法支持传感器模态丢失的训练,这导致了强大的3D目标检测,即使当车辆上的摄像头或雷达传感器突然发生故障时。我们证明了我们的融合方法的有效性,通过广泛的实验辐射数据集,为数不多的大规模数据集,提供雷达射频图像之一。我们的方法的仅摄像头变体在Waymo Open Dataset上的单目3D目标检测中实现了具有竞争力的性能。
Enabling CMF Estimation in Data-Constrained Scenarios: A Semantic-Encoding Knowledge Mining Model
Institute of Transportation Studies, University of California, Davis(加州大学戴维斯分校交通研究所);Department of Civil and Environmental Engineering, Washington State University(华盛顿州立大学土木与环境工程系)