检索结果-内蒙古大学图书馆

LimitAccess: on-device TinyML based robust speech recognition and age classification

Discover Artificial Intelligence 2023年第1期3卷 8页

作者： Maayah, Marina Abunada, Ahlam Al-Janahi, Khawla Ahmed, Muhammad Ejaz Qadir, Junaid Department of Computer Science and Engineering Qatar University Doha Qatar CSIRO’s Data61 Sydney Australia

Automakers from Honda to Lamborghini are incorporating voice interaction technology into their vehicles to improve the user experience and offer value-added services. Speech recognition systems are a key component of smart cars, enhancing convenience and safety for drivers and passengers. In the future, safety-critical features may rely on speech recognition, but this raises concerns about children accessing such services. To address this issue, the LimitAccess system is proposed, which uses TinyML for age classification and helps parents limit children’s access to critical speech recognition services. This study employs a lite convolutional neural network (CNN) model for two different reasons: First, CNN showed superior accuracy compared to other audio classification models for age classification problems. Second, the lite model will be integrated into a microcontroller to meet its limited resource requirements. To train and evaluate our model, we created a dataset that included child and adult voices of the keyword "open". The system approach categorizes voices into age groups (child, adult) and then utilizes that categorization to grant access to a car. The robustness of the model was enhanced by adding a new class (recordings) to the dataset, which enabled our system to detect replay and synthetic voice attacks. If an adult voice is detected, access to start the car will be granted. However, if a child’s voice or a recording is detected, the system will display a warning message that educates the child about the dangers and consequences of the improper use of a car. Arduino Nano 33 BLE sensing was our embedded device of choice for integrating our trained, optimized model. Our system achieved an overall F1 score of 87.7% and 85.89% accuracy. LimitAccess detected replay and synthetic voice attacks with an 88% F1 score. © The Author(s) 2023.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

A Smart Heart Disease Diagnostic System Using Deep Vanilla LSTM

引用

computers, Materials & Continua 2023年第10期77卷 1251-1279页

作者： Maryam Bukhari Sadaf Yasmin Sheneela Naz Mehr Yahya Durrani Mubashir Javaid Jihoon Moon Seungmin Rho Department of Computer Science COMSATS University IslamabadAttock CampusAttockPakistan Department of Computer Science COMSATS University IslamabadIslamabadPakistan Department of Computer Science UET TaxilaTaxilaPakistan Department of AI and Big Data Soonchunhyang UniversityAsanKorea Department of Industrial Security Chung-Ang UniversitySeoul06974Korea

Effective smart healthcare frameworks contain novel and emerging solutions for remote disease diagnostics,which aid in the prevention of several diseases including heart-related *** this context,regular monitoring of cardiac patients through smart healthcare systems based on Electrocardiogram(ECG)signals has the potential to save many *** existing studies,several heart disease diagnostic systems are proposed by employing different state-of-the-art methods,however,improving such methods is always an intriguing area of ***,in this research,a smart healthcare system is proposed for the diagnosis of heart disease using ECG *** proposed framework extracts both linear and time-series information on the ECG signals and fuses them into a single framework *** linear characteristics of ECG signals are extracted by convolution layers followed by Gaussian Error Linear Units(GeLu)and time series characteristics of ECG beats are extracted by Vanilla Long Short-Term Memory Networks(LSTM).Following on,the feature reduction of linear information is done with the help of ID Generalized Gated Pooling(GGP).In addition,data misbalancing issues are also addressed with the help of the Synthetic Minority Oversampling Technique(SMOTE).The performance assessment of the proposed model is done over the two publicly available datasets named MIT-BIH arrhythmia database(MITDB)and PTB Diagnostic ECG database(PTBDB).The proposed framework achieves an average accuracy performance of 99.14%along with a 95%recall value.

关键词： Smart systems deep learning ECG signals heart disease concurrent learning LSTM generalized gated pooling

来源：评论

学校读者我要写书评

暂无评论

Benchmarking Low-Resource Machine Translation Systems 7

Benchmarking Low-Resource Machine Translation Systems

引用

7th Workshop on Technologies for Machine Translation of Low-Resource Languages, LoResMT 2024 at ACL 2024

作者： da Silva, Ana Alexandra Morim Srivastava, Nikit Ngoli, Tatiana Moteu Röder, Michael Moussallem, Diego Ngomo, Axel-Cyrille Ngonga DICE Group Department of Computer Science Paderborn University Germany Jusbrasil Data Science Team Rio de Janeiro Brazil

ISBN: (纸本)9798891761490

Assessing the performance of machine translation systems is of critical value, especially to languages with lower resource availability. Due to the large evaluation effort required by the translation task, studies often compare new systems against single systems or commercial solutions. Consequently, determining the best-performing system for specific languages is often unclear. This work benchmarks publicly available translation systems across 4 datasets and 26 languages, including low-resource languages. We consider both effectiveness and efficiency in our evaluation. Our results are made public through BENG-a FAIR benchmarking platform for Natural Language Generation tasks. © 2024 Association for Computational Linguistics.

关键词： computer aided language translation

来源：评论

学校读者我要写书评

暂无评论

BESP: A novel BCI experiment simulation platform to assist BCI-related algorithm validation 16

BESP: A novel BCI experiment simulation platform to assist B...

引用

16th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, CISP-BMEI 2023

作者： Chen, Qingzhi Ke, Jingshuang Huang, Zhihua Fuzhou University College of Computer and Data Science Fuzhou China

ISBN: (纸本)9798350330755

In recent decades, many brain-computer interface (BCI) software platforms have emerged. However, there are still some limitations. First, integrating an algorithm on online BCI software platform is difficult and time-consuming. Second, there is no guarantee that an unproven algorithm will work properly. Last, existing platforms do not support automatic recording of result metrics during experiments, and frequent validation of multiple algorithms is inefficient. In this paper, we designed a novel BCI experiment simulation platform (BESP) based on offline analysis to satisfy the needs of algorithm validation. BESP reorganizes the experimental steps and provides some interfaces, which makes algorithm validation simple and efficient. In addition, BESP is able to record the results produced during the experimentation. Then the results of multiple algorithms can be easily analyzed, compared and visualized by the statistical analysis system provided by BESP. With these characteristics, BESP is able to verify the correctness of the algorithm. Finally, we designed a simulated P300 spelling experiment based on BESP, and completed the validation and analysis of Algorithms SPLUCB and TSPLUCB (based on PLUCB and TPLUCB). Furthermore, we used BESP to analyze the differences between the stimulus signals of the 20 subjects. Through the analysis, we found some areas worthy of further study. In summary, BESP is well suited for the process of analysis, optimization and validation among multiple algorithms. © 2023 IEEE.

关键词： Simulation platform

来源：评论

学校读者我要写书评

暂无评论

High-Level Synthesis Algorithm for Continuous-Flow Microfiuidic Biochips Under Device Volume Constraints

High-Level Synthesis Algorithm for Continuous-Flow Microfiui...

引用

2023 International Conference on computer science and Automation Technology, CSAT 2023

作者： Chen, Zhengyang Pan, Youlin Chen, Zhen Liu, Genggeng College of Computer and Data Science Fuzhou University Fuzhou China

ISBN: (纸本)9798350371437

Continuous-flow microfluidic biochips (CFMBs) can precisely manipulate microfluids for bioassays. In order to enhance the efficiency of bioassays, it is imperative to carefully consider operation scheduling and device binding schemes. The existing work has considered device volume constraints at high-level synthesis to meet the actual fluid transport requirements of the CFMBs. However, this work is based on integer linear programming, which has a high algorithmic complexity and makes it difficult to meet the large-scale integration requirements of CFMBs. Therefore, a fast high-level synthesis algorithm based on a discrete particle swarm algorithm under device volume constraint is proposed. We propose a particle encoding scheme based on the operation priorities and the device binding scheme. In addition, the corresponding particle initialization method and update strategy are proposed to satisfy the device volume constraint. The experimental results show that the proposed algorithm can obtain high-level synthesis results in a short time. © 2023 IEEE.

关键词： Integer programming

来源：评论

学校读者我要写书评

暂无评论

Traffic Flow Prediction Method Based on Similarity Graph and Temporal Convolution Units

Traffic Flow Prediction Method Based on Similarity Graph and...

引用

2023 International Conference on Smart Systems for Applications in Electrical sciences, ICSSES 2023

作者： Chen, Hao Chen, Chi-Hua College of Computer and Data Science Fuzhou University Fuzhou China

ISBN: (纸本)9798350347296

Accurate traffic flow prediction can improve urban commuting efficiency, but how to effectively mine the spatio-temporal characteristics of traffic data is the biggest challenge. Therefore, this paper proposes a spatio-temporal convolution method, which can effectively extract the traffic dynamic correlation of different road nodes, and reconstruct the road network by cosine similarity principle. This principle can reflect the similar situation of traffic change trend of two road nodes in the past period of time. In order to sample the time-dependent features of road nodes, this study uses a time-dependent convolutional network, which has the ability to extract sequence time-dependent features from longer data. © 2023 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Topological Coherence Preserving Retinal Vessel Segmentation 13

Topological Coherence Preserving Retinal Vessel Segmentation

引用

13th International Conference on Information Technology in Medicine and Education, ITME 2023

作者： Mao, Meiyan Lin, Jiawen Liu, Shilin College of Computer and Data Science Fuzhou University Fuzhou China

ISBN: (纸本)9798350319156

Retinal vessel segmentation is a critical step in computer-aided diagnosis of retinal images, providing rich information of vascular morphology. However, existing works primarily focus on achieving high overall segmentation accuracy, often overlooking the errors in the topological structure that lead to disconnected vessels. In order to address these issues, a topological coherence preserving retinal vessel segmentation method is proposed. We adopt a two-stage model by cascading a refinement step after the typical segmentation network to enhance retinal vessel connectivity. A topology-preserving loss is then introduced to guide the refinement network to pay more attention to key regions with topological differences, allowing it to learn the correct topology structure and maximize the preservation of retinal vessel tree connectivity. Experimental results on the DRIVE, STARE, and CHASE_DB1 datasets demonstrate the effectiveness of the proposed method. © 2023 IEEE.

关键词： Topology

来源：评论

学校读者我要写书评

暂无评论

Joint Task Assignment and Migration in Cloud-Edge-End Collaborative Computing Based on DRL

Joint Task Assignment and Migration in Cloud-Edge-End Collab...

引用

2023 International Conference on Intelligent Communication and Networking, ICN 2023

作者： Zhang, Xiaoqi Cheng, Hongju Fuzhou University College of Computer and Data Science Fuzhou China

ISBN: (纸本)9798350357677

By effectively assigning and migrating tasks based on service requirements, the success rate of task execution in cloud-edge-end collaborative computing can be significantly enhanced, thereby ensuring the provision of high-quality services for users. The majority of conventional cloud-edge-end task offloading approaches primarily focus on static scenarios, posing challenges in ensuring the success rate of task execution in mobile scenarios. It is imperative to address the problem of constructing a joint optimization scheme for task allocation and migration that is suitable for mobile scenarios. This paper re-define the latency, energy, and migration model for task processing in mobile scenarios. Furthermore, we propose a Deep Reinforcement learning (DRL)-based Task allocation and Migration optimization algorithm (DRTM) to enhance the efficiency of task completion and minimize the total cost. DRTM introduces the traditional Actor-Critic with a mirror deep deterministic policy gradient (DDPG) and establishes a duel Q-network to update parameters on respective gradients for optimal policy acquisition. DRTM incorporates two target networks to effectively improve stability and convergence speed during training while reducing computational complexity. The experimental results demonstrate that DRTM can offer a high-performance task assignment and migration scheme in mobile scenarios, thereby significantly reducing the total cost of the task execution life cycle. © 2023 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Graph Convolutional Network with Learnable Message Propagation Mechanism 9

Graph Convolutional Network with Learnable Message Propagati...

引用

9th International Conference on Information, Cybernetics, and Computational Social Systems, ICCSS 2023

作者： Cai, Hai-Chun Lin, Yue-Na Zhang, Chun-Yang School of Computer and Data Science Fuzhou University Fuzhou China

ISBN: (纸本)9798350342239

In recent years, graph neural network has become the main paradigm for solving graph analysis tasks, which can easily process high-dimensional data and has a powerful fitting capability. Recent works on graph neural networks have successfully transferred the convolution network in computer vision to graph. Graph convolution network (GCN) has become the classical network framework in graph neural networks due to its simple aggregation approach and favorable theoretical support. When the original graph data are constructed, an adjacency matrix is used to represent the topology, where 0 or 1 indicates whether there is a connection between nodes. Moreover, GCN aggregates node attributes only depends on adjacency matrix. Although it can learn a mapping function, its message propagation mechanism is fixed for a given adjacency matrix. However, for specific downstream tasks, we expect to propagate messages relevant to the downstream task, while a fixed aggregation mode cannot handle this. To this end, we propose a graph convolution neural network with a learnable message propagation mechanism. The original adjacency matrix is adjusted through a learnable weight, so that the message propagation mechanism better adapts to downstream tasks. Experimental results show that the proposed model achieves significant performance in node classification. © 2023 IEEE.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Uneven Hybrid Clickstream Generation Based on Multi-Layer Perceptron 11

Uneven Hybrid Clickstream Generation Based on Multi-Layer Pe...

引用

11th International Conference on Advanced Cloud and Big data, CBD 2023

作者： Liu, Shen Yu, Zhiyong Huang, Fangwan Fuzhou University College of Computer and Data Science Fuzhou China

ISBN: (纸本)9798350345346

Online social networking is prevalent around the world, meanwhile, social bots damage users' experience by stealing user privacy, sending spam, spreading malicious links, launching DDoS attacks, et al. In this paper, we use social bots to automatically manipulate social accounts based on generated clickstreams to demonstrate the vulnerability of social networking sites and the capability of social bots. Clickstreams are the sequential click events of users or bots browsing a website. Existing research on clickstream generation assumes that the intervals between all click events are equal, which makes it difficult to fully capture the relationship between uneven intervals. To this end, we propose an uneven hybrid clickstream generation algorithm based on multi-layer perceptron (UHCG-MLP), which can generate click events and intervals simultaneously by dealing with uneven intervals. UHCG-MLP firstly transforms each clickstream in the training set into a sequence of discrete elements separated by coded interval, and uses multi-layer perceptron to mine user behavior patterns from these sequences. Then it generates hybrid clickstreams according to the prediction of the trained model. Experiments on the Twitter platform using the clickstreams generated by the proposed model and the other five benchmark models show that UHCG-MLP achieves the best execution effect. © 2023 IEEE.

关键词： Denial-of-service attack

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：