检索结果-内蒙古大学图书馆

Source-free Subject Adaptation for EEG-based Visual Recognition

学校读者我要写书评

暂无评论

Source-free Subject Adaptation for EEG-based Visual Recognit...

International Winter Workshop on Brain-computer Interface (BCI)

作者： Pilhyeon Lee Seogkyu Jeon Sunhee Hwang Minjung Shin Hyeran Byun Department of Computer Science Yonsei University Seoul Republic of Korea Vision AI Team LG Uplus Seoul Republic of Korea Graduate School of Artificial Intelligence Yonsei University Seoul Republic of Korea Graduate School of Artificial Intelligence and Graduate Program of Cognitive Scienc Yonsei University

This paper focuses on subject adaptation for EEG-based visual recognition. It aims at building a visual stimuli recognition system customized for the target subject whose EEG samples are limited, by transferring knowledge from abundant data of source subjects. Existing approaches consider the scenario that samples of source subjects are accessible during training. However, it is often infeasible and problematic to access personal biological data like EEG signals due to privacy issues. In this paper, we introduce a novel and practical problem setup, namely source-free subject adaptation, where the source subject data are unavailable and only the pre-trained model parameters are provided for subject adaptation. To tackle this challenging problem, we propose classifier-based data generation to simulate EEG samples from source subjects using classifier responses. Using the generated samples and target subject data, we perform subject-independent feature learning to exploit the common knowledge shared across different subjects. Notably, our framework is generalizable and can adopt any subject-independent learning method. In the experiments on the EEG-ImageNet40 benchmark, our model brings consistent improvements regardless of the choice of subject-independent learning. Also, our method shows promising performance, recording top-l test accuracy of 74.6% under the 5-shot setting even without relying on source data. Our code can be found at https://***/DeepBCI/Deep-BCI.

关键词： Training Representation learning Visualization Adaptation models Target recognition Brain modeling Electroencephalography

Innovations in Social Informatic: A Patent Landscape Review

学校读者我要写书评

暂无评论

Procedia computer science 2024年 234卷 1228-1235页

作者： Nur Afia Agung Purnomo Fairuz Iqbal Maulana Yogi Tri Prasetyo Ma. Janice J. Gumasing Elsa Rosyidah Satria Fadil Persada Institute for Research and Community Services Universitas Nahdlatul Ulama Sidoarjo Sidoarjo 61234 Indonesia English Language Education Universitas Nahdlatul Ulama Sidoarjo Sidoarjo 61234 Indonesia Entrepreneurship Department BINUS Business School Undergraduate Program Bina Nusantara University Jakarta 11480 Indonesia Computer Science Department Bina Nusantara University Jakarta 11480 Indonesia School of Industrial Engineering and Engineering Management Mapúa University Manila 1002 Philippines International Bachelor Program in Engineering Yuan Ze University Taoyuan 32003 Taiwan

Patent intellectual property in social informatics with entrepreneurship support creates greater opportunities to improve human life. It has the potential to open enormous innovation opportunities in various fields, such as technology, health, services, and society. This research was conducted to review the patent landscape and the main trends of patents on social informatics in all countries. Researchers applied a patent landscape analysis using data on 487 patent documents over 16 years from the *** database. The study results show yearly patented development patterns related to social informatics. There is a royalty-free opportunity by using 15 patent discontinued. Patent related to social informatics was dominated by company ownership and inventors from the United States. The patent physics (G) section overwhelmed patents related to social informatics. There were 334 simple family patents on social informatics in nine jurisdictions; the United States was the highest.

关键词： entrepreneurship, patent, social informatics innovation

Understanding Audiovisual Deepfake Detection: Techniques, Challenges, Human Factors and Perceptual Insights

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hashmi, Ammarah Shahzad, Sahibzada Adil Lin, Chia-Wen Tsao, Yu Wang, Hsin-Min The Social Networks and Human-Centered Computing Program Taiwan International Graduate Program Institute of Information Science Academia Sinica Taipei11529 Taiwan The Institute of Information Systems and Applications National Tsing Hua University Hsinchu300044 Taiwan The Department of Computer Science National Chengchi University Taipei11529 Taiwan The Department of Electrical Engineering The Institute of Communications Engineering National Tsing Hua University Hsinchu300044 Taiwan The Research Center for Information Technology Innovation Academia Sinica Taipei11529 Taiwan The Institute of Information Science Academia Sinica Taipei11529 Taiwan

Deep Learning has been successfully applied in diverse fields, and its impact on deepfake detection is no exception. Deepfakes are fake yet realistic synthetic content that can be used deceitfully for political impersonation, phishing, slandering, or spreading misinformation. Despite extensive research on unimodal deepfake detection, identifying complex deepfakes through joint analysis of audio and visual streams remains relatively unexplored. To fill this gap, this survey first provides an overview of audiovisual deepfake generation techniques, applications, and their consequences, and then provides a comprehensive review of state-of-the-art methods that combine audio and visual modalities to enhance detection accuracy, summarizing and critically analyzing their strengths and limitations. Furthermore, we discuss existing open source datasets for a deeper understanding, which can contribute to the research community and provide necessary information to beginners who want to analyze deep learning-based audiovisual methods for video forensics. By bridging the gap between unimodal and multimodal approaches, this paper aims to improve the effectiveness of deepfake detection strategies and guide future research in cybersecurity and media integrity. © 2024, CC BY.

关键词： Audiovisual

AGAINST THE LEANS: OVERCOMING SPATIAL DISORIENTATION THROUGH GALVANIC VESTIBULAR STIMULATION 65th

学校读者我要写书评

暂无评论

AGAINST THE LEANS: OVERCOMING SPATIAL DISORIENTATION THROUGH...

65th Human Factors and Ergonomics Society Annual Meeting, HFES 2021

作者： Kim, Sungho Lazaro, May Jorella Jung, Hyunki Yun, Myung Hwan Kang, Yohan Department of Industrial Engineering Seoul National University Seoul Korea Republic of Interdisciplinary Program in Cognitive Science Seoul National University Seoul Korea Republic of Beyond Honeycomb Seoul Korea Republic of Republic of Korea Air Force Gyeryong Korea Republic of

ISBN: (纸本)9781713844129

Leans illusion is a type of Spatial Disorientation (SD) that pilots often experience which can adversely affect flight performance. For pilots' flight safety, research on how to effectively overcome SD such as leans illusion is important. The purpose of this study is to identify the overcoming effect of Galvanic Vestibular Stimulation (GVS) technology on leans illusion. Twenty-one Air Force pilots participated in a flight simulation experiment where leans illusion was induced through a specialized SD simulator. In the with-GVS condition, GVS was given during the roll-out phase. Data was analyzed using roll angle error and subjective SD scales by two conditions (with-GVS, without-GVS). Results showed that both the roll angle error and the subjective SD scale scores were found to be lower in the with-GVS condition than in the without-GVS condition. This study suggests that the use of GVS technology can potentially contribute in overcoming leans illusion. © 2021 by Human Factors and Ergonomics Society. All rights reserved.

关键词： Flight simulators

Advances in classifying the stages of diabetic retinopathy using convolutional neural networks in low memory edge devices

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Paul, Aditya Jyoti Cognitive Applications Research Lab India Department of Computer Science and Engineering SRM Institute of Science and Technology Tamil Nadu Kattankulathur603203 India

Diabetic Retinopathy (DR) is a severe complication that may lead to retinal vascular damage and is one of the leading causes of vision impairment and blindness. DR broadly is classified into two stages - non-proliferative (NPDR), where there are almost no symptoms, except a few microaneurysms, and proliferative (PDR) involving a huge number of microaneurysms and hemorrhages, soft and hard exudates, neo-vascularization, macular ischemia or a combination of these, making it easier to detect. More specifically, DR is usually classified into five levels, labeled 0-4, from 0 indicating no DR to 4 which is most severe. This paper firstly presents a discussion on the risk factors of the disease, then surveys the recent literature on the topic followed by examining certain techniques which were found to be highly effective in improving the prognosis accuracy. Finally, a convolutional neural network model is proposed to detect all the stages of DR on a low-memory edge microcontroller. The model has a size of just 5.9 MB, accuracy and F1 score both of 94% and an inference speed of about 20 frames per *** Codes 68T45, 68T10, 68T07, 68U10 Copyright © 2021, The Authors. All rights reserved.

关键词： Eye protection

Strategies of Collaborative Filtering Recommendation Mechanism Using a Deep Learning Approach 3

学校读者我要写书评

暂无评论

Strategies of Collaborative Filtering Recommendation Mechani...

3rd IEEE Eurasia Conference on IOT, Communication and engineering, ECICE 2021

作者： Chang, Chin-Chih Chang, Huan-Ming Chung Hua University Becholar Program in Applied Foreign Language and Smart Commerce Hsinchu30012 Taiwan Chung Hua University Department of Computer Science and Information Engineering Hsinchu30012 Taiwan

ISBN: (纸本)9781665445160

Nowadays, recommendation systems are widely used to help users locate the items they want. Collaborative filtering (CF) is a commonly used method for the recommendation. CF techniques use user-item ratings for prediction but suffer from the problems of data sparsity, cold start, and scalability. Though the Matrix Factorization (MF) techniques like Singular Value Decomposition (SVD) or Principal Component Analysis (PCA) overcome the above-mentioned problems, these methods are possible to deliver unmeaningful results in the condition of a low ranked approximation and denser singular vectors. In this paper, we review strategies of collaborative filtering recommendation mechanisms and propose an approach based on an autoencoder of convolutional neural network. Autoencoders are unsupervised learning methods in which neural networks are supported for the task of representation learning. We identify the user's features through learning, and then use these features to combine the collaborative filtering algorithm to recommend items. The experimental results show that the convolutional autoencoder can effectively reduce the computations when the amount of data is huge and benefited from the performance of its convolutional neural network. © 2021 IEEE.

关键词： Recommender systems

Unmasking Illusions: Understanding Human Perception of Audiovisual Deepfakes

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hashmi, Ammarah Shahzad, Sahibzada Adil Lin, Chia-Wen Tsao, Yu Wang, Hsin-Min Social Networks and Human-Centered Computing Program Taiwan International Graduate Program Institute of Information Science Academia Sinica Taipei11529 Taiwan Institute of Information Systems and Applications National Tsing Hua University Hsinchu300044 Taiwan Department of Computer Science National Chengchi University Taipei11529 Taiwan Department of Electrical Engineering Institute of Communications Engineering National Tsing Hua University Hsinchu300044 Taiwan Research Center for Information Technology Innovation Academia Sinica Taipei11529 Taiwan Institute of Information Science Academia Sinica Taipei11529 Taiwan

The emergence of contemporary deepfakes has attracted significant attention in machine learning research, as artificial intelligence (AI) generated synthetic media increases the incidence of misinterpretation and is difficult to distinguish from genuine content. Currently, machine learning techniques have been extensively studied for automatically detecting deepfakes. However, human perception has been less explored. Malicious deepfakes could ultimately cause public and social problems. Can we humans correctly perceive the authenticity of the content of the videos we watch? The answer is obviously uncertain;therefore, this paper aims to evaluate the human ability to discern deepfake videos through a subjective study. We present our findings by comparing human observers to five state-of-the-art deepfake detection models. To this end, we used gamification concepts to provide 110 participants (55 native English speakers and 55 non-native English speakers) with a web-based platform where they could access a series of 40 videos (20 real and 20 fake) to determine their authenticity. Each participant performed the experiment twice with the same 40 videos in different random orders. The videos are manually selected from the FakeAVCeleb dataset. We found that all AI models performed significantly better than humans when evaluated on the same 40 videos. The study also reveals that while deception is not impossible, humans tend to overestimate their detection capabilities. Our experimental results may help benchmark human versus machine performance, advance forensics analysis, and enable adaptive countermeasures. © 2024, CC BY.

关键词： Audiovisual

Augmenting a Description Logic with Probability for Motion Patterns Within QSTR 6th

学校读者我要写书评

暂无评论

Augmenting a Description Logic with Probability for Motion ...

6th International Conference on Advanced Computing, Networking, and Informatics, ICACNI 2018

作者： Talukdar, Upasana Barua, Rupam Hazarika, Shyamanta M. Biomimetic & Cognitive Robotics Lab Department of Computer Science & Engineering Tezpur University Tezpur India Department of Computer Science & Engineering Jorhat Engineering College Jorhat India Mechatronics and Robotics Lab Department of Mechanical Engineering Indian Institute of Technology Guwahati India

ISBN: (纸本)9789811396793

Motion patterns are spatiotemporal. Human everyday spatiotemporal reasoning is predominantly qualitative;hence, a qualitative abstraction of motion patterns holds promise. Reasoning over qualitative models of spatiotemporal change is referred to as qualitative spatiotemporal reasoning (QSTR). Recognition of patterns within a QSTR framework involves reasoning over a multitude of entities. These entities interacting in numerous ways often lead to hierarchies and ample uncertainties. Herein lies the motivation of using a formal basis for the representation of large and diverse knowledge. Description logics (DLs), a family of knowledge representation formalism providing object-oriented representation with formal semantics, are a natural choice. However, traditional DLs delineate a delicate approach to handle uncertain features and relationships. A paradigm that combines a DL formalization with probability theory is required. Bayesian Networks (BNs) are the most widely used models to represent knowledge about an uncertain domain. This paper combines a DL with BN and proposes a formal, explicit qualitative knowledge representation formalism for recognition of motion patterns within an uncertain domain. © 2020, Springer Nature Singapore Pte Ltd.

关键词： Bayesian networks

A comparative study of transformer-based language models on extractive question answering

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Pearce, Kate Zhan, Tiffany Komanduri, Aneesh Zhan, Justin Department of Computer Science and Computer Engineering University of Arkansas Army Educational Outreach Program UNITE

Question Answering (QA) is a task in natural language processing that has seen considerable growth after the advent of transformers. There has been a surge in QA datasets that have been proposed to challenge natural language processing models to improve human and existing model performance. Many pre-trained language models have proven to be incredibly effective at the task of extractive question answering. However, generalizability remains as a challenge for the majority of these models. That is, some datasets require models to reason more than others. In this paper, we train various pre-trained language models and fine-tune them on multiple question answering datasets of varying levels of difficulty to determine which of the models are capable of generalizing the most comprehensively across different datasets. Further, we propose a new architecture, BERT-BiLSTM, and compare it with other language models to determine if adding more bidirectionality can improve model performance. Using the F1-score as our metric, we find that the RoBERTa and BART pre-trained models perform the best across all datasets and that our BERT-BiLSTM model outperforms the baseline BERT model. © 2021, CC BY.

关键词： Natural language processing systems