检索结果-内蒙古大学图书馆

Mitigating Misleading Chain-of-Thought Reasoning with Selective Filtering

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wu, Yexin Zhang, Zhuosheng Zhao, Hai College of Zhiyuan Shanghai Jiao Tong University China School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University China Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University China

Large language models have manifested remarkable capabilities by leveraging chain-of-thought (CoT) reasoning techniques to solve intricate questions through step-by-step reasoning chains. Despite its success, the efficacy of such reasoning is inherently contingent upon the quality of CoT. However, flawless CoT reasoning cannot be guaranteed due to the presence of indecomposable questions and the potential for erroneous reasoning chains, particularly in the case of small-scale language models. To tackle this challenge, we propose a novel approach called the selective filtering reasoner (SelF-Reasoner) that assesses the entailment relationship between the question and the candidate reasoning chain. Then, we proceed with CoT reasoning when the reasoning chain demonstrates confidence;otherwise, we opt to predict the answer directly. SelF-Reasoner improves the fine-tuned T5 baseline consistently over the scienceQA, ECQA, and LastLetter tasks. Code is available at https://***/LibroWu/SelF-Reasoner. © 2024, CC BY-SA.

关键词： Computational linguistics

Parallel self-organization map using multiple stimuli

学校读者我要写书评

暂无评论

Parallel self-organization map using multiple stimuli

International Joint Conference on Neural Networks (IJCNN)

作者： M. Yasunaga K. Tominaga Jung Hwan Kim Institute of Information Sciences and Electronics University of Tsukuba Tsukuba Japan Master Program of Science and Engineering University of Tsukuba Tsukuba Japan Center for Advanced Computer Studied University of Southwestern Louisiana Lafayette LA USA

We propose a parallel SOM algorithm to speedup the fundamental SOM calculation using parallel computer environments. In the parallel SOM algorithm synaptic weights are updated in parallel corresponding to multiple stimuli (inputs). Parallelism in the proposed algorithm is based on the analogy of the biological neural networks in which neurons respond to the stimuli in parallel. Performance is evaluated by implementing the newly developed performance simulator in a personal computer-cluster under the message passing interface library (MPI) environment. A speedup ratio of about 5.0 is achieved with 8 processors (personal computers) when the width of the neighborhood function is less than 5% of the total number of neurons in the SOM network.

关键词： Biology computing Neurons Concurrent computing Parallel processing Biological neural networks Biological system modeling Computational modeling computer simulation computer interfaces Message passing

Individual-based simulation of sexual selection: A quantitative genetic approach

学校读者我要写书评

暂无评论

Procedia computer science 2010年第1期1卷 2003-2011页

作者： D. van Dijk P.M.A. Sloot J.C. Tay M.C. Schut Computational Science University of Amsterdam Sciencepark 107 1098 XG Amsterdam The Netherlands Evolutionary and Complex Systems Program School of Computer Engineering Nanyang Technological University Singapore 639798 Singapore Department of Artificial Intelligence Vrije Universiteit Amsterdam 1081 HV Amsterdam The Netherlands

Sexual selection has been mathematically modeled using quantitative genetics as well as population genetics. Two-locus simulation models have been used to study the evolution of male display and female preference. We present an individual-based simulation model of sexual selection in a quantitative genetic context. We show that under certain conditions Fisherian self-reinforcing sexual selection takes effect, predicted by Lande’s analytic model of female choice. We also show that the dynamics involved in the co-evolution of male display and female preference is much more complex than mathematics would predict. We therefore argue that the study of sexual selection through individual-based simulation could give new and more realistic insight into a world dominated by overly simplified equations.

关键词： Sexual selection Individual-based modeling Agent-based modeling

Information and motor constraints shape melodic diversity across cultures

学校读者我要写书评

暂无评论

arXiv 2024年

作者： McBride, John M. Kim, Nahie Nishikawa, Yuri Saadakeev, Mekhmed Pearce, Marcus Tlusty, Tsvi Center for Algorithmic and Robotized Synthesis Institute for Basic Science Ulsan44919 Korea Republic of School of Business Administration Ulsan National Institute of Science and Technology Ulsan44919 Korea Republic of Department of Molecular Life Science Tokai University School of Medicine Kanagawa Japan Department of Biomedical Engineering Ulsan National Institute of Science and Technology Ulsan44919 Korea Republic of Cognitive Science Research Group School of Electronic Engineering & Computer Science Queen Mary University of London London United Kingdom Department of Clinical Medicine Aarhus University Aarhus Denmark Center for Soft and Living Matter Institute for Basic Science Ulsan44919 Korea Republic of Departments of Physics and Chemistry Ulsan National Institute of Science and Technology Ulsan44919 Korea Republic of

The number of possible melodies is unfathomably large, yet despite this virtually unlimited potential for melodic variation, melodies from different societies can be surprisingly similar. The motor constraint hypothesis accounts for certain similarities, such as scalar motion and contour shape, but not for other major common features, such as repetition, song length, and scale size. Here we investigate the role of information constraints arising from limitations on human memory in shaping these hallmarks of melodies. We measure determinants of information rate in 62 corpora of Folk melodies spanning several continents, finding multiple trade-offs that all act to constrain the information rate across societies. By contrast, 39 corpora of Art music from Europe (including Turkey) show longer, more complex melodies, and increased complexity over time, suggesting different cultural-evolutionary selection pressures in Art and Folk music, possibly due to the use of written versus oral transmission. Our parameter-free model predicts the empirical scale degree distribution using information constraints on scalar motion, melody length, and, most importantly, information rate. This provides strong evidence that information constraints during cultural transmission of music limit the number of notes in a scale, and proposes that preference for intermediate melodic complexity is a fundamental constraint on the cultural evolution of melody. © 2024, CC BY-NC-SA.

关键词：

Multi-UAVs communication for payload data link enhancement under asynchronous access

学校读者我要写书评

暂无评论

Multi-UAVs communication for payload data link enhancement u...

International Conference on Wireless and Optical Communications ( WOCC)

作者： Yun-Ruei Li Hsin-Piao Lin Li-Chun Wang Abebe Belay Adege Li-Yong Chen Department of Electronic Engineering National Taipei University of Technology Taipei Taiwan Department of Electrical Engineering National Chiao Tung University Hsinchu Taiwan Department of International Graduate Program in Electrical Engineering and Computer Science (EECS) National Taipei University of Technology Taipei Taiwan

Recently, unmanned Aerial Vehicles (UAVs) are widely applied in civilian and commercial areas. In order to rapidly arrange the UAVs in a variable and changeable environment, the technique of maintaining stable and reliable communication link between Ground Control Station (GCS) and UAVs is one of the main issues [1]. However, UAVs deployment can significantly increase the efficiency of missions execution and the flexibility of task allocation. As discussed by different scholars, multi-UAVs have a number of advantages depending on the application areas: disaster areas monitoring, wireless sensor data collection in precision agriculture, and relay-aided of UAVs network topology [2].

关键词： OFDM Resource management Interference Wireless communication Payloads Bandwidth Optical filters

Borneo Smart Forest Information System for Management of Dipterocarp Plants in Kalimantan Rainforest

学校读者我要写书评

暂无评论

Borneo Smart Forest Information System for Management of Dip...

Information Technology International Seminar (ITIS)

作者： Masna Wati Novianti Puspitasari Ummul Hairah Joan Angelina Widians Anindita Septiarini Ade Fiqri Tjikoa Dept. of Informatics Faculty of Engineering Universitas Mulawarman Samarinda Indonesia Dept. of Informatics Universitas Mulawarman Doctoral Program Dept. of Computer Science and Electronics Universitas Gadjah Mada Yogyakarta Indonesia

ISBN: (纸本)9798350398205

Kalimantan’s tropical rainforest is home to indigenous plants of the Dipterocarpaceae family (tribe), which is renowned for having the most endemic species. Dipterocarp is a family of pantropical plants, many of which are used in the wood industry. This study aims to provide information on the potential of Dipterocarp in Kalimantan forests. Education and conservation efforts for Dipterocarp trees, a prominent component of the forest ecosystem, are crucial in light of the significant harm done to Kalimantan’s rainforests. The authors develop the Borneo Smart Forest information system. This system digitally organizes data on the diversity of Dipterocarp in the Kalimantan rainforest based on Web and QR-Code technology. Borneo Smart Forest (BSF) system for monitoring plant inventory and plant markers can be read by smartphones using QR codes. It gives visitors a comprehensive overview of the plants. A QR code is a printed, two-dimensional barcode with limited space for data storage. This technology allows the plant to show earlier administrator-made additions to the inventory information system. The QR code will direct the user to the BSF website, which has data about the plants. The development of this system is one of the efforts to support the achievement of Dipterocarp conservation.

关键词： Seminars Memory Forestry QR codes Vegetation Wood industry Information technology

Sub-structure Characteristic Mode Analysis of Microstrip Antennas Using a Global Multi-trace Formulation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Ran Lu, Yuyu Cheng, Guang Shang Zhu, Wei Hu, Jun Bagci, Hakan Chengdu611731 China The Electrical and Computer Engineering Program Computer Electrical and Mathematical Science and Engineering Division King Abdullah University of Science and Technology Thuwal23955 Saudi Arabia The Information Materials and Intelligent Sensing Laboratory of Anhui Province Anhui University Hefei230601 China The Anhui Province Key Laboratory of Target Recognition and Feature Extraction Lu’an237010 China Thuwal23955 Saudi Arabia

A characteristic mode (CM) method that relies on a global multi-trace formulation (MTF) of surface integral equations is proposed to compute the modes and the resonance frequencies of microstrip patch antennas with finite dielectric substrates and ground planes. Compared to the coupled formulation of electric field and Poggio-Miller-Chang-Harrington-Wu-Tsai integral equations, global MTF allows for more direct implementation of a sub-structure CM method. This is achieved by representing the coupling of the electromagnetic fields on the substrate and ground plane in the form of a numerical Green function matrix, which yields a more compact generalized eigenvalue equation. The resulting sub-structure CM method avoids the cumbersome computation of the multilayered medium Green function (unlike the CM methods that rely on mixed-potential integral equations) and the volumetric discretization of the substrate (unlike the CM methods that rely on volume-surface integral equations), and numerical results show that it is a reliable and accurate approach to predicting the modal behavior of electromagnetic fields on practical microstrip antennas. Copyright © 2022, The Authors. All rights reserved.

关键词： Microstrip antennas

Defending pre-trained language models from adversarial word substitutions without performance sacrifice

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Bao, Rongzhou Wang, Jiayi Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Shanghai China

Pre-trained contextualized language models (PrLMs) have led to strong performance gains in downstream natural language understanding tasks. However, PrLMs can still be easily fooled by adversarial word substitution, which is one of the most challenging textual adversarial attack methods. Existing defence approaches suffer from notable performance loss and complexities. Thus, this paper presents a compact and performance-preserved framework, Anomaly Detection with Frequency- Aware Randomization (ADFAR). In detail, we design an auxiliary anomaly detection classifier and adopt a multi-task learning procedure, by which PrLMs are able to distinguish adversarial input samples. Then, in order to defend adversarial word substitution, a frequency-aware randomization process is applied to those recognized adversarial input samples. Empirical results show that ADFAR significantly outperforms those newly proposed defense methods over various tasks with much higher inference speed. Remarkably, ADFAR does not impair the overall performance of PrLMs. © 2021, CC BY.

关键词： Anomaly detection

Pre-training universal language representation

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Yian Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University China

Despite the well-developed cut-edge representation learning for language, most language representation models usually focus on specific levels of linguistic units. This work introduces universal language representation learning, i.e., embeddings of different levels of linguistic units or text with quite diverse lengths in a uniform vector space. We propose the training objective MiSAD that utilizes meaningful n-grams extracted from large unlabeled corpus by a simple but effective algorithm for pre-trained language models. Then we empirically verify that well designed pre-training scheme may effectively yield universal language representation, which will bring great convenience when handling multiple layers of linguistic objects in a unified way. Especially, our model achieves the highest accuracy on analogy tasks in different language levels and significantly improves the performance on downstream tasks in the GLUE benchmark and a question answering dataset. Copyright © 2021, The Authors. All rights reserved.

关键词： Linguistics