检索结果-内蒙古大学图书馆

Outlier Robust Adversarial Training

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hu, Shu Yang, Zhenhuan Wang, Xin Ying, Yiming Lyu, Siwei Department of Computer and Information Technology Purdue School of Engineering and Technology Indiana University-Purdue University Indianapolis IndianapolisIN46202 United States Etsy Inc BrooklynNY United States Department of Epidemiology and Biostatistics School of Public Health University at Albany State University of New York AlbanyNY12222 United States Department of Mathematics and Statistics University at Albany State University of New York AlbanyNY12222 United States Department of Computer Science and Engineering University at Buffalo State University of New York BuffaloNY14260-2500 United States

Supervised learning models are challenged by the intrinsic complexities of training data such as outliers and minority subpopulations and intentional attacks at inference time with adversarial samples. While traditional robust learning methods and the recent adversarial training approaches are designed to handle each of the two challenges, to date, no work has been done to develop models that are robust with regard to the low-quality training data and the potential adversarial attack at inference time simultaneously. It is for this reason that we introduce Outlier Robust Adversarial Training (ORAT) in this work. ORAT is based on a bi-level optimization formulation of adversarial training with a robust rank-based loss function. Theoretically, we show that the learning objective of ORAT satisfies the H-consistency (Awasthi et al., 2021) in binary classification, which establishes it as a proper surrogate to adversarial 0/1 loss. Furthermore, we analyze its generalization ability and provide uniform convergence rates in high probability. ORAT can be optimized with a simple algorithm. Experimental evaluations on three benchmark datasets demonstrate the effectiveness and robustness of ORAT in handling outliers and adversarial attacks. Our code is available at https://***/discovershu/ORAT. Copyright © 2023, The Authors. All rights reserved.

关键词： Statistics

Student and AI responses to physics problems examined through the lenses of sensemaking and mechanistic reasoning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Sirnoorkar, Amogh Zollman, Dean Laverty, James T. Magana, Alejandra J. Rebello, Sanjay Bryan, Lynn A. Purdue University 100 N University St West LafayetteIN47907 United States Department of Physics Kansas State University Cardwell Hall 1228 N Martin Luther King Junior Drive ManhattanKS66506 United States Computer and Information Technology Purdue University 401 Grant Street West LafayetteIN47907 United States Department of Physics and Astronomy Purdue University 525 Northwestern Avenue West LafayetteIN47907 United States Department of Curriculum and Instruction Purdue University 100 N University St West LafayetteIN47907 United States

Several reports in education have called for transforming physics learning environments by promoting sensemaking of real-world scenarios in light of curricular ideas. Recent advancements in Generative-Artificial Intelligence has garnered increasing traction in educators' community by virtue of its potential in transforming STEM learning. In this exploratory study, we adopt a mixed-methods approach in comparatively examining student- and AI-generated responses to two different formats of a physics problem through the cognitive lenses of sensemaking and mechanistic reasoning. The student data is derived from think-aloud interviews of introductory students and the AI data comes from ChatGPT's solutions collected using Zero shot approach. The results highlight AI responses to evidence most features of the two processes through well-structured solutions and student responses to effectively leverage representations in their solutions through iterative refinement of arguments. In other words, while AI responses reflect how physics is talked about, the student responses reflect how physics is practiced. Implications of these results in light of development and deployment of AI systems in physics pedagogy are discussed. © 2023, CC BY.

关键词： Students

DynaCon: Dynamic Robot Planner with Contextual Awareness via LLMs

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Kim, Gyeongmin Kim, Taehyeon Kannan, Shyam Sundar Venkatesh, Vishnunandan L.N. Kim, Donghan Min, Byung-Cheol Helper Lab Department of Intelligent Robotics SungKyunKwan University Suwon16419 Korea Republic of SMART Lab Department of Computer and Information Technology Purdue University West LafayetteIN47907 United States HRI Lab Department of Electrical Engineering Kyung Hee University Yongin17104 Korea Republic of

Mobile robots often rely on pre-existing maps for effective path planning and navigation. However, when these maps are unavailable, particularly in unfamiliar environments, a different approach become essential. This paper introduces DynaCon, a novel system designed to provide mobile robots with contextual awareness and dynamic adaptability during navigation, eliminating the reliance of traditional maps. DynaCon integrates real-time feedback with an object server, prompt engineering, and navigation modules. By harnessing the capabilities of Large Language Models (LLMs), DynaCon not only understands patterns within given numeric series but also excels at categorizing objects into matched spaces. This facilitates dynamic path planner imbued with contextual awareness. We validated the effectiveness of DynaCon through an experiment where a robot successfully navigated to its goal using reasoning. Source code and experiment videos for this work can be found at: https://***/view/dynacon. © 2023, CC BY.

关键词： Navigation

Studying Battery Range and Range Anxiety for Electric Vehicles based on Real Travel Demands

学校读者我要写书评

暂无评论

Proceedings of the Human Factors and Ergonomics Society Annual Meeting 2021年第1期65卷 332 - 336页

作者： Zhengming Zhang Renran Tian School of Industrial Engineering Purdue University West Lafayette Indiana USA Department of Computer Information Technology IUPUI Indianapolis Indiana USA

Determination of appropriate battery ranges is critical for developing and utilizing electric cars, which remains an active research topic. In particular, the issues of range anxiety have not been well studied concerning the battery design. Towards these research gaps, this study firstly investigates the baseline battery ranges based on the actual travel data collected from a large-scale longitudinal naturalistic driving study in the Midwestern USA. The occurrences and severity levels of range anxiety are then studied given the baseline, which leads to an augmented optimization model to eliminate such issues. Results show that in the baseline model, 60% of drivers can replace their gas cars entirely with 400-mile battery ranges, and less than 40% can do so with 200-mile battery ranges. Even when all the travel needs are satisfied, the optimal battery ranges can still cause range anxiety issues for all the drivers. An additional 25 miles of battery range can help solve the problem based on the improved optimization results.

关键词：

Establishing Wireless Intranet Network Using UAVs and Web Application for Emergency Communications

学校读者我要写书评

暂无评论

Establishing Wireless Intranet Network Using UAVs and Web Ap...

IEEE Sensors Applications Symposium, SAS

作者： Jihoon Yang Juhyun Kim Sion Kang Haeeun Ok Yijun Yoo Hyewon Koh Pilkyo Kim Alexander L. Head Anthony Smith Dept. of Computer Education Jeju National University Jeju Republic of Korea Dept. of Management Information Systems Jeju National University Jeju Republic of Korea Dept. of Computer Science and Engineering Chungnam National University Daejeon Republic of Korea Dept. of Computer Engineering Jeju National University Jeju Republic of Korea Major of Housing & Family Welfare(Faculty of Human Ecology & Welfare) Jeju National University Jeju Republic of Korea Dept. of Computer Science & Statistics Jeju National University Jeju Republic of Korea Dept. of Computer and Information Technology Purdue University West Lafayette Indiana USA

Every year, natural disasters such as earthquakes, hurricanes, volcanic eruptions, and tsunamis occur around the world. Natural disasters not only cause economic damage and casualties, but also destroy entire infrastructures. When communication infrastructure is damaged, information exchange between rescuers and victims or among rescuers themselves is not smooth, which inevitably reduces the effectiveness of relief activities. The purpose of this study is to establish a communication network using unmanned aerial vehicles (UAVs) known as drones, and to provide smooth communication services to areas where communication facilities have been destroyed through web application development. A low-altitude aerial base station (ABS) is built using a drone and provides a wireless intranet communication network. The intranet-only web application is developed to provide a service that allows people to send and receive messages and location information. In future research, separating the communication network between rescuers and non-rescuers and using multiple ABSs to build the network could serve more people in a wider disaster area.

关键词：

Federated Optimization of Smooth Loss Functions

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jadbabaie, Ali Makur, Anuran Shah, Devavrat The Laboratory for Information and Decision Systems Massachusetts Institute of Technology CambridgeMA02139 United States The Department of Computer Science The Elmore Family School of Electrical and Computer Engineering Purdue University West LafayetteIN47907 United States

In this work, we study empirical risk minimization (ERM) within a federated learning framework, where a central server seeks to minimize an ERM objective function using n samples of training data that is stored across m clients and the server. The recent flurry of research in this area has identified the Federated Averaging (FedAve) algorithm as the staple for determining ϵ-approximate solutions to the ERM problem. Similar to standard optimization algorithms, e.g., stochastic gradient descent, the convergence analysis of FedAve and its variants only relies on smoothness of the loss function in the optimization parameter. However, loss functions are often very smooth in the training data too. To exploit this additional smoothness in data in a federated learning context, we propose the Federated Low Rank Gradient Descent (FedLRGD) algorithm. Since smoothness in data induces an approximate low rank structure on the gradient of the loss function, our algorithm first performs a few rounds of communication between the server and clients to learn weights that the server can use to approximate clients' gradients using its own gradients. Then, our algorithm solves the ERM problem at the server using an inexact gradient descent method. To theoretically demonstrate that FedLRGD can have superior performance to FedAve, we present a notion of federated oracle complexity as a counterpart to canonical oracle complexity in the optimization literature. Under some assumptions on the loss function, e.g., strong convexity and smoothness in the parameter, η-Hölder class smoothness in the data, etc., we prove that the federated oracle complexity of FedLRGD scales like (Equation presented) and that of FedAve scales like (Equation presented) (neglecting typically sub-dominant factors), where 1 is the ratio of client-to-server communication time to gradient computation time, p is the parameter dimension, and d is the data dimension. Then, we show that when d is small compared to n and the l

关键词： Optimization

Peek into the Future Camera-based Occupant Sensing in Configurable Cabins for Autonomous Vehicles

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Prabu, Avinash Tian, Renran Li, Lingxi Le, Jialiang Sundararajan, Srinivasan Barbat, Saeed The Department of Electrical & Computer Engineering Indiana University Purdue University Indianapolis 723 West Michigan Street IndianapolisIN United States The Department of Computer Information Technology Indiana University Purdue University Indianapolis 799 West Michigan Street IndianapolisIN United States The Ford Motor Company Research and Innovation Center DearbornMI United States

The development of fully autonomous vehicles (AVs) can potentially eliminate drivers and introduce unprecedented seating design. However, highly flexible seat configurations may lead to occupants' unconventional poses and actions. Understanding occupant behaviors and prioritize safety features become eye-catching topics in the AV research frontier. Visual sensors have the advantages of cost-efficiency and high-fidelity imaging and become more widely applied for in-car sensing purposes. Occlusion is one big concern for this type of system in crowded car cabins. It is important but largely unknown about how a visual-sensing framework will look like to support 2-D and 3-D human pose tracking towards highly configurable seats. As one of the first studies to touch this topic, we peek into the future camera-based sensing framework via a simulation experiment. Constructed representative car-cabin, seat layouts, and occupant sizes, camera coverage from different angles and positions is simulated and calculated. The comprehensive coverage data are synthesized through an optimization process to determine the camera layout and overall occupant coverage. The results show the needs and design of a different number of cameras to fully or partially cover all the occupants with changeable configurations of up to six seats. © 2022, CC BY.

关键词： Cameras

Energy Efficient Collaborative Federated Learning Design: A Graph Neural Network based Approach

学校读者我要写书评

暂无评论

Energy Efficient Collaborative Federated Learning Design: A ...

IEEE Conference on Global Communications (GLOBECOM)

作者： Nuocheng Yang Sihua Wang Mingzhe Chen Christopher G. Brinton Changchuan Yin Beijing Laboratory of Advanced Information Network Beijing University of Posts and Telecommunications Beijing China State Key Laboratory of Networking and Switching Technology Beijing University of Posts and Telecommunications Beijing China Department of Electrical and Computer Engineering Institute for Data Science and Computing University of Miami Coral Gables FL USA School of Electrical and Computer Engineering Purdue University West Lafayette IN USA

In this paper, we consider the design of an energy efficient collaborative federated learning (CFL) methodology where devices exchange their local FL parameters with a subset of their neighbors without reliance on a parameter server. In the considered model, mobile devices implement the designed CFL to train their local FL models using their own datasets over a realistic wireless network. Due to the limited wireless resources and user movements, each device may not be able to transmit its FL parameters with all neighboring devices. Therefore, each device must select a subset of devices to share its FL parameters and optimize the transmit power. This problem is formulated as an optimization problem, whose goal is to minimize CFL training energy consumption while satisfying the delay and CFL training loss requirements. To solve this problem, a two-stage solution is proposed. At the first stage, a graph neural network (GNN) based algorithm is proposed, which enables each device to individually determine the subset of devices to transmit FL parameters using its neighboring devices' location and connection information. Compared to standard iterative algorithms that need to iteratively optimize device connections and transmit power, the proposed GNN based method can directly obtain the optimal device connections without iterative optimization. Given the optimal device connections, at the second stage, each device can directly obtain the optimal transmit power. Simulation results show that the proposed algorithm can decrease energy consumption by up to 46% compared to the algorithm where each device will directly connect to its first and second nearest neighbors.

关键词：

Do language models learn typicality judgments from text?

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Misra, Kanishka Ettinger, Allyson Rayz, Julia Taylor Department of Computer and Information Technology Purdue University IN United States Department of Linguistics University of Chicago IL United States

Building on research arguing for the possibility of conceptual and categorical knowledge acquisition through statistics contained in language, we evaluate predictive language models (LMs)-informed solely by textual input-on a prevalent phenomenon in cognitive science: typicality. Inspired by experiments that involve language processing and show robust typicality effects in humans, we propose two tests for LMs. Our first test targets whether typicality modulates LM probabilities in assigning taxonomic category memberships to items. The second test investigates sensitivities to typicality in LMs' probabilities when extending new information about items to their categories. Both tests show modest-but not completely absent-correspondence between LMs and humans, suggesting that text-based exposure alone is insufficient to acquire typicality knowledge. © 2021, CC BY.

关键词： Knowledge representation