检索结果-内蒙古大学图书馆

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Campbell, Declan Rane, Sunayana Giallanza, Tyler De Sabbata, Nicolò Ghods, Kia Joshi, Amogh Ku, Alexander Frankland, Steven M. Griffiths, Thomas L. Cohen, Jonathan D. Webb, Taylor Princeton Neuroscience Institute United States Department of Psychology Princeton University United States Department of Computer Science EPFL Switzerland Department of Cognitive Science Dartmouth College United States Department of Computer Science Princeton University United States Microsoft Research United States

Recent work has documented striking heterogeneity in the performance of state-of-the-art vision language models (VLMs), including both multimodal language models and text-to-image models. These models are able to describe and generate a diverse array of complex, naturalistic images, yet they exhibit surprising failures on basic multi-object reasoning tasks - such as counting, localization, and simple forms of visual analogy - that humans perform with near perfect accuracy. To better understand this puzzling pattern of successes and failures, we turn to theoretical accounts of the binding problem in cognitive science and neuroscience, a fundamental problem that arises when a shared set of representational resources must be used to represent distinct entities (e.g., to represent multiple objects in an image), necessitating the use of serial processing to avoid interference. We find that many of the puzzling failures of state-of-the-art VLMs can be explained as arising due to the binding problem, and that these failure modes are strikingly similar to the limitations exhibited by rapid, feedforward processing in the human brain. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Fine-Tuning Fasttext Using Bayesian Optimization for Movie Review Sentiment Analysis 1

Fine-Tuning Fasttext Using Bayesian Optimization for Movie R...

引用

1st International Conference on Technology, Engineering, and Computing Applications, ICTECA 2023

作者： Halim, Rico Girsang, Abba Suganda Riftiarrasyid, Mohammad Faisal Binus Graduate Program Computer Science Bina Nusantara University Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350360554

The advancements in technology during the 20th century resulted in the onset of the digital computer era. This study investigates the relative importance of earlier algorithms in comparison to more recent ones. Various research projects suggest improving vectorization by combining conventional and modern techniques, while others suggest optimizing it through algorithmic methods. This study primarily focuses on employing Bayesian optimization to optimize hyperparameters, hence improving the performance evaluation of TF-IDF FastText sentiment classification models. This study presented four models: the initial model utilized FastText with a preprocessing step that involved removing stopwords, the second model employed Fast-Text without any optimization using Bayesian optimization, the third model utilized FastText with optimization using Bayesian optimization, and the fourth model combined FastText with TF-IDF and was further optimized using Bayesian optimization. The Support Vector Machine (SVM) technique will be employed to evaluate all of the models. The findings suggest that the model's performance stays unchanged when stopwords are eliminated (Precision, Recall, F1: 0.9007). The model demonstrates a slight enhancement through the utilization of Bayesian optimization, resulting in a 0.02% increase, reaching a final accuracy of 0.9019%. Bayesian optimization is frequently employed to improve performance. The performance of Bayesian optimization is affected by the magnitude of the learning rate, and models that do not utilize Bayesian optimization demonstrate inferior performance. Both the TF-IDF FasText model and FastText model yielded comparable outcomes, attaining an F1 score of 0.9019 after being tuned by Bayesian optimization. © 2023 IEEE.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

Object Size Recognition as Intra-class Variations using Transfer Learning

Object Size Recognition as Intra-class Variations using Tran...

引用

2023 International Conference on computer science, Information Technology and Engineering, ICCoSITE 2023

作者： Rahmania, Rissa Hendric Spits Warnars, Harco Leslie Soewito, Benfano Gaol, Ford Lumban Bina Nusantara University BINUS Graduate Program Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350320954

The ability to differentiate various products in the retail store plays an essential role to provide effectiveness to customers and reduce or even eliminate long queues. However, traditional machine learning algorithms are incapable of recognizing many subordinate categories in various retail product. This paper aims to recognize the retail product recognition algorithm based on the YOLOv7 model in terms of intra-class variations, with the sub-categories of brand and size. We used two schemes of the dataset to compare recognition performance between them. Firstly, the YOLOv7 is applied in the two schemes of the dataset that is annotated with the subordinate category to detect the brand as meta category. Secondly, the proposed method is applied by adding the object size classification into the YOLOv7 model where the square area of the bounding box was calculated to classify the product according to size. Confidence score and square area are used to verify the object and to obtain the product size, which represents sub-category of the product. The experimental results show that our proposed method achieves higher recall compared to baseline object detection. © 2023 IEEE.

关键词： Object recognition

来源：评论

学校读者我要写书评

暂无评论

Machine Learning-Based Crop Recommendation for IoT-Enabled Smart Agriculture 24

Machine Learning-Based Crop Recommendation for IoT-Enabled S...

引用

9th International Conference on Cloud Computing and Internet of Things, CCIOT 2024

作者： Sonata, Ilvico Computer Science Department Binus Graduate Program Bina Nusantara University Indonesia Jakarta Indonesia

ISBN: (纸本)9798400717161

Human food needs are increasing all the time. This condition is in line with the increasing world population and exacerbated by climate change. Agricultural products are one of the sources of food to meet food needs. The use of technology is one of the keys to success in increasing agricultural production. One of the technologies used in agricultural technology is the Internet of Things (IoT). IoT in agricultural technology can be used to monitor soil and weather conditions to ensure that these conditions are in accordance with the needs of agricultural crops. The use of machine learning technology can be combined with IoT to help predict the types of crops that are suitable for planting based on soil and weather conditions. The results of the experiment show that the machine learning model can be used to predict the right types of plants to plant based on soil and weather conditions obtained through IoT. This method can be used to increase agricultural yields. Copyright © 2024 held by the owner/author(s). Publication rights licensed to ACM.

关键词： Smart agriculture

来源：评论

学校读者我要写书评

暂无评论

Abstractive Indonesian News Summarization Using BERT2GPT 7

Abstractive Indonesian News Summarization Using BERT2GPT

引用

7th International Conference on Information Technology, Information Systems and Electrical Engineering, ICITISEE 2023

作者： Nasari, Maulin Maulina, Ayu Girsang, Abba Suganda Computer Science Bina Nusantara University Binus Graduate Program Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350382266

This study examines the necessity of employing BERT2GPT for single-document summarization in the current age of escalating digital data. The primary focus of this work is on the abstractive technique, which tries to generate versatile summaries that surpass the primary sentences of the original content. This study employs two models, namely BERT and GPT -2, for the purpose of the summarization system. The paper introduces the BERT2GPT model, which merges the bidirectional characteristics of BERT with the generative powers of GPT. The findings indicate that the BERT2GPT method successfully catches significant information and linguistic nuances, hence enhancing the quality of the generated summaries. The corresponding average values for Rl, R2, and RL are 0.62, 0.56, and 0.60, respectively. © 2023 IEEE.

关键词： Abstractive Summarization BERT2GPT single-document summarization

来源：评论

学校读者我要写书评

暂无评论

Transformer and CNN Comparison for Time Series Classification Model 15

Transformer and CNN Comparison for Time Series Classificatio...

引用

15th International Congress on Advanced Applied Informatics Winter, IIAI-AAI-Winter 2023

作者： Sonata, Ilvico Heryadi, Yaya Binus Graduate Program Doctor of Computer Science Bina Nusantara University Computer Science Department Jakarta Indonesia

ISBN: (纸本)9798350383829

In real life, many activities are performed sequentially. These activities must be carried out sequentially, such as the assembly process in the manufacturing production process. This series of activities cannot be reduced or added so that the main goal of the series of activities is achieved. Apart from that, there are also time series events that occur naturally, such as rainy and hot conditions in a certain area. The classification process of time series activities is very important to see the possibility of anomalies occurring. The significant development of machine learning models in recent years has made the process of classifying time series data increasingly researched. Several previous studies stated that deep learning models were more accurate in classifying time series data. In this paper, we will compare Convolutional Neural Network (CNN) and Transformer deep learning models in classifying time series data. Experimental results using the same public datasets for CNN and Transformer model show that the CNN model is more accurate than the Transformer model. The results of measuring accuracy using confusion matrix show that CNN has an accuracy of 92% and Transformer has an accuracy of 80%. © 2023 IEEE.

关键词： Time series

来源：评论

学校读者我要写书评

暂无评论

Trade Ease with Machine Learning and AWS

引用

IEEE Access 2023年 11卷 25893-25905页

作者： Teja, Kamurthi Ravi Liu, Chuan-Ming National Taipei University of Technology International Program of Electrical Engineering and Computer Science Taipei10608 Taiwan National Taipei University of Technology Department of Computer Science and Information Engineering Taipei10608 Taiwan

Global trading is undergoing significant changes, necessitating modifications to the trading strategies. This study presents a newly developed cloud-based trading strategy that uses Amazon Web Services (AWS), machine learning (ML), and data science to automate trading tasks. The study begins by creating a machine learning trading strategy using a customized deep neural network (DNN). The strategy was then tested using a trading station before deployment on a cloud machine. The execution of the strategy resulted in a total profit of 42.74526 USD, with a test accuracy score of 87.45% and a training accuracy score of 89.15% over 6047 epochs. The roadmap provides a step-by-step overview of the entire process from strategy development to execution. In addition, this study offers insights into related issues and solutions that can enhance the effectiveness of trading strategies. Overall, this study contributes significantly to the field of cloud-based trading strategies and opens avenues for future research. © 2013 IEEE.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Review on Machine Learning Methods for Water Quality Prediction 4

A Review on Machine Learning Methods for Water Quality Predi...

引用

4th International Conference on Artificial Intelligence and Data sciences, AiDAS 2023

作者： Dharma, Vickenzio Leeputra Nurtanio, Nathanael Kenneth Nugroho, Fakhru Sidik Anggreainy, Maria Susan Kurniawan, Afdhal School of Computer Science Bina Nusantara University Computer Science Department Jakarta11480 Indonesia Bina Nusantara University Binus Graduate Program Doctor of Computer Science Computer Science Department Jakarta11530 Indonesia School of Computer Science Bina Nusantara University Computer Science Department Jakarta Indonesia

ISBN: (纸本)9798350318432

Water is an important substance for the human body. Clean water is important for not just the human body, but also for the environment. In this paper, Prisma is used to filter many of the reference paper where the paper left are used for the research. Machine learning is one of many ways to predict water quality. Using algorithms to process large amounts of data and patterns, water quality can be predicted. Prediction on water temperature, pH, and others can be used to find bad quality water and potentially find the cure for it. There are also different kinds of environment in predicting water quality, where depending on the environment certain machine learning method is better than the others. This paper contains machine learning methods, water parameters, and areas of water for water quality prediction. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

User Guide for KOTE: Korean Online That-gul Emotions Dataset 30

User Guide for KOTE: Korean Online That-gul Emotions Dataset

引用

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Jeon, Duyoung Lee, Junho Kim, Cheongtag Department of Psychology Seoul National University Korea Republic of Interdisciplinary Program in Cognitive Science Seoul National University Korea Republic of

ISBN: (纸本)9782493814104

Despite the lack of comprehensive exploration of emotional connotations, sentiment analysis, which categorizes data as positive or negative, has been widely employed to identify emotional aspects in texts. Recently, corpora labeled with more than just valence or polarity have been built to surpass this limitation. However, most Korean emotion corpora are limited by their small size and narrow range of emotions covered. In this paper, we introduce the KOTE dataset. The KOTE dataset comprises 50,000 Korean online comments, totaling 250,000 cases, each manually labeled for 43 emotions and NO EMOTION through crowdsourcing. The taxonomy for the 43 emotions was systematically derived through cluster analysis of Korean emotion concepts within the word embedding space. After detailing the development of KOTE, we further discuss the results of fine-tuning, as well as analysis for social discrimination within the corpus. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

Software Quality Assessment Methods and Standards in Weld Defect Detection for Shipbuilding 6

Software Quality Assessment Methods and Standards in Weld De...

引用

6th International Conference on Cybernetics and Intelligent System, ICORIS 2024

作者： Pernando, Yonky Gaol, Ford Lumban Soeparno, Haryono Arifin, Yulyani Computer Science Department BINUS Graduate Program Bina Nusantara University Jakarta Indonesia

ISBN: (纸本)9798331539603

Shipbuilder safety and performance, structural integrity of welded joints in shipbuilding is of utmost importance. Traditional manual inspection methods to detect any potential defect can be time consuming and susceptible to human errors;alternative inspection systems have proven more cost effective and accurate solutions than manual approaches for defect identification. This paper investigates how advanced software quality assessment standards and machine learning techniques such as Convolutional Neural Networks (CNNs) and You Only Look Once (YOLO) object identification framework can enhance weld defect detection. Research undertaken for this thesis addresses three core questions. 1) Which standards provide effective software quality assessments used for weld defect detection systems?;and 2) In what ways will they impact efficiency and reliability. How will implementation of these standards increase accuracy in manufacturing processes? Integrating high-quality annotated datasets, optimizing joint preparation and parameter settings, and using cloud solutions for computing needs can enable industries to achieve more reliable, efficient, and scalable defect detection. This proactive approach not only enhances immediate product quality, but also contributes to long-term durability and performance, decreasing maintenance costs and increasing overall reliability for steel plates used for marine applications. Furthermore, evidence has revealed that adopting software quality assessment standards in weld defect detection significantly enhances structural integrity as well as operational efficiency of shipbuilding operations. © 2024 IEEE.

关键词： Convolutional neural networks

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：