检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Rakshit, Geetanjali Flanigan, Jeffrey Computer Science and Engineering Department UC Santa Cruz United States

Figurative language is commonplace in natural language, and while making communication memorable and creative, can be difficult to understand. In this work, we investigate the robustness of Question Answering (QA) models on figurative text. Yes/no questions, in particular, are a useful probe of figurative language understanding capabilities of large language models. We propose FigurativeQA, a set of 1000 yes/no questions with figurative and non-figurative contexts, extracted from the domains of restaurant and product reviews. We show that state-of-the-art BERT-based QA models exhibit an average performance drop of up to 15% points when answering questions from figurative contexts, as compared to non-figurative ones. While models like GPT-3 and ChatGPT are better at handling figurative texts, we show that further performance gains can be achieved by automatically simplifying the figurative contexts into their non-figurative (literal) counterparts. We find that the best overall model is ChatGPT with chain-of-thought prompting to generate non-figurative contexts. Our work provides a promising direction for building more robust QA models with figurative language understanding capabilities. © 2023, CC BY-NC-SA.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficient In-Training Adaptive Compound Loss Function Contribution Control for Medical Image Segmentation 46

Efficient In-Training Adaptive Compound Loss Function Contri...

引用

46th Annual International Conference of the IEEE engineering in Medicine and Biology Society, EMBC 2024

作者： Al-Battal, Abdullah F. Duong, Soan T. M. Nguyen, Chanh D. Tr. Truong, Steven Q. H. Phan, Chien Nguyen, Truong Q. An, Cheolhong Uc San Diego Electrical and Computer Engineering Department La Jolla CA United States King Fahd University of Petroleum and Minerals Electrical Engineering Department Saudi Arabia Vinbrain Jsc Hanoi Viet Nam Le Quy Don Technical University Department of Computer Science Hanoi Viet Nam VinUniversity College of Engineering and Computer Science Hanoi Viet Nam University Medical Center Diagnostic Imaging Department Ho Chi Minh City Viet Nam

ISBN: (纸本)9798350371499

Image segmentation plays a crucial role in many clinical applications, including disease diagnosis and monitoring. Current state-of-the-art segmentation approaches use deep neural networks that are trained on their target tasks by minimizing a loss function. Class imbalance is one of the major challenges that these networks face, where the target object is significantly underrepresented. Compound loss functions that incorporate the binary cross-entropy (BCE) and Dice loss are among the most prominent approaches to address this issue. However, determining the contribution of each individual loss to the overall compound loss function is a tedious process. It requires hyperparameter fine-tuning and multiple iterations of training, which is highly inefficient in terms of time and energy consumption. To address this issue, we propose an approach that adaptively controls the contribution of each of these individual loss functions during training. This eliminates the need for multiple fine-tuning iterations to achieve the desired precision and recall for segmentation models. © 2024 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Interactions beyond the Pandemic: Lessons Learned from Large-scale Emergency Remote Teaching in Higher Education 25

Interactions beyond the Pandemic: Lessons Learned from Large...

引用

2025 CHI Conference on Human Factors in Computing Systems, CHI 2025

作者： Yarmand, Matin Li, Haowei Weibel, Nadir Computer Science and Engineering The Design Lab UC San Diego San Diego CA United States Graduate School of Education University of Pennsylvania Philadelphia PA United States

ISBN: (纸本)9798400713941

Online education - given the enhanced access for diverse populations and flexible participation - has been a topic of interest for many computer science and learning science researchers. The sudden shift to online settings during the COVID-19 Emergency Remote Teaching (ERT) provided a valuable opportunity to examine the use of educational technologies on a global scale with various digital readiness skills, beyond many past works that relied on small lab studies. Following a PRISMA-inspired methodology grounded on Moore's three types of classroom interaction, this descriptive review investigates 22 empirical research papers published during the COVID-19 ERT era focused on higher-education online classrooms. We explore the empirical evidence reported in the collected corpus, and given how ERT remains a likely future occurrence, we suggest key directions for future research, including a new learning paradigm that centralizes and augments Learner-Content interaction to balance between flexibility and structure of online learning. © 2025 Copyright held by the owner/author(s).

关键词： Emergency Remote Education Higher Education Learning Interactions Online Classroom

来源：评论

学校读者我要写书评

暂无评论

Learning Stabilization Control from Observations by Learning Lyapunov-like Proxy Models

arXiv

引用

arXiv 2023年

作者： Ganai, Milan Hirayama, Chiaki Chang, Ya-Chien Gao, Sicun The Department of Computer Science and Engineering UC San Diego United States

The deployment of Reinforcement Learning to robotics applications faces the difficulty of reward engineering. Therefore, approaches have focused on creating reward functions by Learning from Observations (LfO) which is the task of learning policies from expert trajectories that only contain state sequences. We propose new methods for LfO for the important class of continuous control problems of learning to stabilize, by introducing intermediate proxy models acting as reward functions between the expert and the agent policy based on Lyapunov stability theory. Our LfO training process consists of two steps. The first step attempts to learn a Lyapunov-like landscape proxy model from expert state sequences without access to any kinematics model, and the second step uses the learned landscape model to guide in training the learner's policy. We formulate novel learning objectives for the two steps that are important for overall training success. We evaluate our methods in real automobile robot environments and other simulated stabilization control problems in model-free settings, like Quadrotor control and maintaining upright positions of Hopper in MuJoCo. We compare with state-of-the-art approaches and show the proposed methods can learn efficiently with less expert observations. Copyright © 2023, The Authors. All rights reserved.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Powering an E-Ink Display from Soil Bacteria 21

Powering an E-Ink Display from Soil Bacteria

引用

19th ACM Conference on Embedded Networked Sensor Systems, SenSys 2021

作者： Marcano, Gabriel Pannuto, Pat UC San Diego Computer Science and Engineering United States

ISBN: (纸本)9781450390972

This demo showcases the power delivery potential of soil-based microbial fuel cells. We build a prototype energy harvesting setup for a soil microbial fuel cell, measure the amount of power that we can harvest, and use that energy to drive an e-ink display. Microbial fuel cells are highly sensitive to environmental conditions, especially soil moisture. In near-optimal, super moist conditions our cell provides approximately 100 μW of power at around 500 mV, which is ample power over time to power our system several times a day. In sum, we find that the confluence of ever lower-power electronics and new understanding of microbial fuel cell design means that "soil-powered sensors"are now feasible. There remains, however, significant future work to make these systems reliable and maximally performant. This demo is a working copy of the system presented at LP-IoT'21 [6]. © 2021 Owner/Author.

关键词： Microbial fuel cells

来源：评论

学校读者我要写书评

暂无评论

Data Redaction from Pre-trained GANs

Data Redaction from Pre-trained GANs

引用

Secure and Trustworthy Machine Learning (SaTML), IEEE Conference on

作者： Zhifeng Kong Kamalika Chaudhuri Computer Science and Engineering UC San Diego La Jolla USA

Large pre-trained generative models are known to occasionally output undesirable samples, which undermines their trustworthiness. The common way to mitigate this is to re-train them differently from scratch using different data or different regularization - which uses a lot of computational resources and does not always fully address the problem. In this work, we take a different, more compute- friendly approach and investigate how to post-edit a model after training so that it “redacts”, or refrains from outputting certain kinds of samples. We show that redaction is a fundamentally different task from data deletion, and data deletion may not always lead to redaction. We then consider Generative Adversar-ial Networks (GANs), and provide three different algorithms for data redaction that differ on how the samples to be redacted are described. Extensive evaluations on real-world image datasets show that our algorithms out-perform data deletion baselines, and are capable of redacting data while retaining high generation quality at a fraction of the cost of full re- training,

关键词：

来源：评论

学校读者我要写书评

暂无评论

Appliancizer: Transformingweb pages into electronic devices 21

Appliancizer: Transformingweb pages into electronic devices

引用

2021 CHI Conference on Human Factors in Computing Systems: Making Waves, Combining Strengths, CHI 2021

作者： Garza, Jorge Merrill, Devon J. Swanson, Steven Computer Science and Engineering Uc San Diego United States

ISBN: (纸本)9781450380966

Prototyping electronic devices that meet today's consumer standards is a time-consuming task that requires multi-domain expertise. Consumers expect electronic devices to have visually appealing interfaces with both tactile and screen-based interfaces. Appliancizer, our interactive computational design tool, exploits the similarities between graphical and tangible interfaces, allowing web pages to be rapidly transformed into physical electronic devices. Using a novel technique we call essential interface mapping, our tool converts graphical user interface elements (e.g., an HTML button) into tangible interface components (e.g., a physical button) without changing the application source code. Appliancizer automatically generates the PCB and low-level code from web-based prototypes and HTML mock-ups. This makes the prototyping of mixed graphical-tangible interactions as easy as modifying a web page and allows designers to leverage the well-developed ecosystem of web technologies. We demonstrate how our technique simplifes and accelerates prototyping by developing two devices with Appliancizer. © 2021 ACM.

关键词： Graphical user interfaces

来源：评论

学校读者我要写书评

暂无评论

Soil Power?: Can Microbial Fuel Cells Power Non-Trivial Sensors? 1

Soil Power?: Can Microbial Fuel Cells Power Non-Trivial Sens...

引用

1st ACM Workshop on No Power and Low Power Internet-of-Things, LP-IoT 2021 - Part of MOBICOM 2021

作者： Marcano, Gabriel Pannuto, Pat Uc San Diego Computer Science and Engineering United States

ISBN: (纸本)9781450387026

This paper explores the power delivery potential of soil-based microbial fuel cells. We build a prototype energy harvesting setup for a soil microbial fuel cell, measure the amount of power that we can harvest, and use that energy to drive an e-ink display as a representative example of a periodic energy-intensive load. Microbial fuel cells are highly sensitive to environmental conditions, especially soil moisture. In near-optimal, super moist conditions our cell provides approximately 100 μW of power at around 500 mV, which is ample power over time to power our system several times a day. We further explore how cell performance diminishes and recovers with varying moisture levels as well as how cell performance is affected by the load from the energy harvester itself. In sum, we find that the confluence of ever lower-power electronics and new understanding of microbial fuel cell design means that "soil-powered sensors"are now feasible. There remains, however, significant future work to make these systems reliable and maximally performant. © 2022 Owner/Author.

关键词： Energy harvesting

来源：评论

学校读者我要写书评

暂无评论

SCALE-UP: AN EFFICIENT BLACK-BOX INPUT-LEVEL BACKDOOR DETECTION VIA ANALYZING SCALED PREDICTION CONSISTENCY 11

SCALE-UP: AN EFFICIENT BLACK-BOX INPUT-LEVEL BACKDOOR DETECT...

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Guo, Junfeng Li, Yiming Chen, Xun Guo, Hanqing Sun, Lichao Liu, Cong Department of Computer Science UT Dallas United States Tsinghua Shenzhen International Graduate School Tsinghua University China Samsung Research America Mountain View United States Department of Computer Science Michigan State University United States Department of Computer Science Lehigh University United States Department of Electricity and Computer Engineering Department UC Riverside United States

Deep neural networks (DNNs) are vulnerable to backdoor attacks, where adversaries embed a hidden backdoor trigger during the training process for malicious prediction manipulation. These attacks pose great threats to the applications of DNNs under the real-world machine learning as a service (MLaaS) setting, where the deployed model is fully black-box while the users can only query and obtain its predictions. Currently, there are many existing defenses to reduce backdoor threats. However, almost all of them cannot be adopted in MLaaS scenarios since they require getting access to or even modifying the suspicious models. In this paper, we propose a simple yet effective black-box input-level backdoor detection, called SCALE-UP, which requires only the predicted labels to alleviate this problem. Specifically, we identify and filter malicious testing samples by analyzing their prediction consistency during the pixel-wise amplification process. Our defense is motivated by an intriguing observation (dubbed scaled prediction consistency) that the predictions of poisoned samples are significantly more consistent compared to those of benign ones when amplifying all pixel values. Besides, we also provide theoretical foundations to explain this phenomenon. Extensive experiments are conducted on benchmark datasets, verifying the effectiveness and efficiency of our defense and its resistance to potential adaptive attacks. Our codes are available at https://***/JunfengGo/SCALE-UP. © 2023 11th International Conference on Learning Representations, ICLR 2023. All rights reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Near-Tight Bounds for 3-Query Locally Correctable Binary Linear Codes via Rainbow Cycles

arXiv

引用

arXiv 2024年

作者： Alrabiah, Omar Guruswami, Venkatesan Department of Electrical Engineering and Computer Science UC Berkeley BerkeleyCA94709 United States Department of Electrical Engineering and Computer Science Department of Mathematics the Simons Institute the Theory of Computing UC Berkeley BerkeleyCA94709 United States

We prove that a binary linear code of block length n that is locally correctable with 3 queries against a fraction δ > 0 of adversarial errors must have dimension at most Oδ(log2 n · log log n). This is almost tight in view of quadratic Reed-Muller codes being a 3-query locally correctable code (LCC) with dimension Θ(log2 n). Our result improves, for the binary field case, the Oδ(log8 n) bound obtained in the recent breakthrough of [KM23a] (and the more recent improvement to Oδ(log4 n) for binary linear codes announced in [Yan24]). Previous bounds for 3-query linear LCCs proceed by constructing a 2-query locally decodable code (LDC) from the 3-query linear LCC/LDC and applying the strong bounds known for the former. Our approach is more direct and proceeds by bounding the covering radius of the dual code, borrowing inspiration from [IS20]. That is, we show that if x 7→ (v1·x, v2·x, . . ., vn·x) is an arbitrary encoding map Fk2 → Fn2 for the 3-query LCC, then all vectors in Fk2 can be written as a Oeδ(log n)-sparse linear combination of the vi’s, which immediately implies k 6 Oeδ((log n)2). The proof of this fact proceeds by iteratively reducing the size of any arbitrary linear combination of at least Ωeδ(log n) of the vi’s. We achieve this using the recent breakthrough result of [ABS+23] on the existence of rainbow cycles in properly edge-colored graphs, applied to graphs capturing the linear dependencies underlying the local correction property. © 2024, CC BY.

关键词： Forward error correction

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：