检索结果-内蒙古大学图书馆

8th IEEE International Conference on Data Science and Advanced Analytics (DSAA)

作者： Parolin, Erick Skorupa Khan, Latifur Osorio, Javier Brandt, Patrick T. D'Orazio, Vito Holmes, Jennifer Univ Texas Dallas Dept Comp Sci Richardson TX 75083 USA Univ Texas Dallas Sch Econ Polit & Policy Sci Richardson TX 75083 USA Univ Texas Dallas Sch Govt & Publ Policy Richardson TX 75083 USA

ISBN: (纸本)9781665420990

Political scientists and security agencies increasingly rely on computerized event data generation to track conflict processes and violence around the world. However, most of these approaches rely on pattern-matching techniques constrained by large dictionaries that are too costly to develop, update, or expand to emerging domains or additional languages. In this paper, we provide an effective solution to those challenges. Here we develop the 3M-Transformers (Multilingual, Multi-label, Multi-task) approach for event coding from domain specific multilingual corpora, dispensing external large repositories for such task, and expanding the substantive focus of analysis to organized crime, an emerging concern for security research. Our results indicate that our 3M-Transformers configurations outperform state-of-the-art usual Transformers models (BERT and XLM-RoBERTa) for coding events on actors, actions and locations in English, Spanish, and Portuguese languages.

关键词： event coding transfer learning natural language processing organized crime deep neural networks multi-task learning

来源：评论

学校读者我要写书评

暂无评论

An event coding Method Based on Frame Images With Dynamic Vision Sensor Modeling

An Event Coding Method Based on Frame Images With Dynamic Vi...

引用

Signal, Information and Data Processing (ICSIDP), IEEE International Conference on

作者： Xingpeng Lu Mingyu Wang Yang Cao Ruilian Zhao Wu Zhou Zhaolin Li School of Information Science and Technology Tsinghua University Beijing China Beijing Aerospace Chenxin Technology Co. Ltd. Beijing China College of Information Science and Technology Beijing University of Chemical Technology Beijing China

ISBN: (数字)9781728123455

ISBN: (纸本)9781728123462

The dynamic vision sensor (DVS) has shown great application potential in the field of real-time object detections due to the advantages of low latency, high dynamic range and low memory usage. DVS utilizes the event-driven principles to extract the features of moving objects and records feature information as the address-event (AE) data. Prior work has shown that DVS-based spike neural networks (SNNs) have obvious advantages on energy efficiency by converting the AE data to spike events. However, most of the prior work mainly relies on specific DVS cameras, which limits the applications of DVS-based SNNs. Then, based on the principles of DVS, this paper presents an event coding method based on frame images. This method generates triggered self-defined AE (SAE) data by using differential comparisons among frame images. The SAE data can be converted to spike events. Compared with pixel-based spike coding methods, the features of moving objects can be represented more efficiently by the proposed method, and the redundant spikes are reduced for low power considerations.

关键词： DVS modeling event coding frame image SNN

来源：评论

学校读者我要写书评

暂无评论

Near Real-Time Atrocity event coding 14

Near Real-Time Atrocity Event Coding

引用

14th IEEE International Conference on Intelligence and Security Informatics - Cybersecurity and Big Data (IEEE ISI)

作者： Solaimani, Mohiuddin Salam, Sayeed Mustafa, Ahmad M. Khan, Latifur Brandt, Patrick T. Thuraisingham, Bhavani Univ Texas Dallas Dept Comp Sci Richardson TX 75083 USA Univ Texas Dallas Sch Econ Polit & Policy Sci Richardson TX 75083 USA

ISBN: (纸本)9781509038657

In recent years, mass atrocities, terrorism, and political unrest have caused much human suffering. Thousands of innocent lives have been lost to these events. With the help of advanced technologies, we can now dream of a tool that uses machine learning and natural language processing (NLP) techniques to warn of such events. Detecting atrocities demands structured event data that contain metadata, with multiple fields and values (e.g. event date, victim, perpetrator). Traditionally, humans apply common sense and encode events from news stories but this process is slow, expensive, and ambiguous. To accelerate it, we use machine coding to generate an encoded event. In this paper, we develop a near-real-time supervised machine coding technique with an external knowledge base, WordNet, to generate a structured event. We design a Spark-based distributed framework with a web scraper to gather news reports periodically, process, and generate events. We use Spark to reduce the performance bottleneck while processing raw text news using CoreNLP.

关键词： Natural language processing Stream mining Incremental learning Atrocity event Apache Spark event coding Big Data

来源：评论

学校读者我要写书评

暂无评论

Multi-CoPED: A Multilingual Multi-Task Approach for coding Political event Data on Conflict and Mediation Domain 22

Multi-CoPED: A Multilingual Multi-Task Approach for Coding P...

引用

AAAI/ACM Conference on AI, Ethics, and Society (AIES)

作者： Parolin, Erick Skorupa Hosseini, MohammadSaleh Hu, Yibo Khan, Latifur Brandt, Patrick T. Osorio, Javier D'Orazio, Vito Univ Texas Dallas Richardson TX 75080 USA Univ Arizona Tucson AZ USA

ISBN: (纸本)9781450392471

Political and social scientists monitor, analyze and predict political unrest and violence, preventing (or mitigating) harm, and promoting the management of global conflict. They do so using event coder systems, which extract structured representations from news articles to design forecast models and event-driven continuous monitoring systems. Existing methods rely on expensive manual annotated dictionaries and do not support multilingual settings. To advance the global conflict management, we propose a novel model, Multi-CoPED (Multilingual Multi-Task Learning BERT for coding Political event Data), by exploiting multi-task learning and state-of-the-art language models for coding multilingual political events. This eliminates the need for expensive dictionaries by leveraging BERT models' contextual knowledge through transfer learning. The multilingual experiments demonstrate the superiority of Multi-CoPED over existing event coders, improving the absolute macro-averaged F1-scores by 23.3% and 30.7% for coding events in English and Spanish corpus, respectively. We believe that such expressive performance improvements can help to reduce harms to people at risk of violence.

关键词： artificial intelligence and geopolitics political conflict social conflict event coding natural language processing transfer learning

来源：评论

学校读者我要写书评

暂无评论

A Lossless Compression Algorithm with Hardware Implementation for Dynamic Vision Sensor

A Lossless Compression Algorithm with Hardware Implementatio...

引用

IEEE International Symposium on Circuits and Systems (ISCAS)

作者： Ding, Zewei Wang, Shangmei Cm, Yujie Zeng, Xiaoyang Li, Wenhong Wang, Mingyu Fudan Univ State Key Lab ASIC & Syst Shanghai 200433 Peoples R China

ISBN: (纸本)9798350330991;9798350331004

Nowadays, with the increase resolution of Dynamic Vision Sensor (DVS), efficient compression algorithm for event stream is needed urgently. Conventional DVS system encodes event data in address event representation (AER) for output while ignores the data redundancy imposed by the correlation of events. To address this challenge, this paper first analyzes the spatiotemporal characteristics of event stream and the impact of readout circuits. Based on the analysis, the context-based encoding strategies for spatial address, timestamp and polarity of events are proposed respectively with the consideration of data flow in DVS hardware. Besides, the hardware architecture with high parallelism is presented to implement the compression algorithm, which achieves high throughput at an affordable cost. The hardware is implemented in the 55nm process as part of a 512x512 resolution DVS. The experimental results demonstrate that our methods achieves higher average compression ratio compared to conventional and DVS-specific coding algorithms.

关键词： Dynamic Vision Sensor (DVS) event coding spatiotemporal correlation context-based coding

来源：评论

学校读者我要写书评

暂无评论

Study of foveation mechanisms in Dynamic Vision Sensors 38

Study of foveation mechanisms in Dynamic Vision Sensors

引用

38th Conference on Design of Circuits and Integrated Systems (DCIS)

作者： Ortiz-Ramirez, Isabel Camunas-Mesa, Luis A. Linares-Barranco, Bernabe Serrano-Gotarredona, Teresa US Inst Microelect Sevilla CSIC Seville Spain

ISBN: (纸本)9798350303858

This paper presents an algorithm to emulate in software the foveation mechanism to be implemented by an electronically foveated dynamic vision sensor. The algorithm has been applied for the foveation of recordings of Dynamic Vision Sensors observing dynamic scenes. The reduction in the number of events is demonstrated while preserving the original information in the detected regions of interest.

关键词： dynamic vision sensor event coding foveation

来源：评论

学校读者我要写书评

暂无评论

event coding and motor priming: how attentional modulation may influence binding across action properties

引用

EXPERIMENTAL BRAIN RESEARCH 2012年第1期219卷 139-150页

作者： Ocampo, Brenda Painter, David R. Kritikos, Ada Univ Queensland Sch Psychol St Lucia Qld 4072 Australia Univ Queensland Queensland Brain Inst St Lucia Qld 4072 Australia

When we perceive an action it is internally transformed into a motor representation akin to the execution of that same action. Motor priming studies show that action observation facilitates the execution of physically similar actions, but interferes with the performance of dissimilar actions. Some evidence suggests, however, that once a specific motor plan is formed, perceiving an action with partially overlapping features (e.g. a congruent grip-type but and incongruent end-goal) can result in interference. In two experiments we investigate how modulating attention towards observed actions influences the binding that occurs between action features, and therefore the amount of partial-overlap interference to participants' performance. In the first study we directed attention towards a salient action feature (the grip-type). We found that perceiving partially overlapping (i.e. partially congruent) actions slowed participants' responses compared with observation of completely congruent or incongruent actions. In the second experiment attentional resources were taxed through the use of a secondary task. This resulted in the elimination of the partial-overlap interference effect. We discuss results in relation to feature binding and event codes.

关键词： Action observation Motor priming Kinematics event coding

来源：评论

学校读者我要写书评

暂无评论

A Fast Track Approach Towards Automatic Detection of Gray Zone Activities from Text

A Fast Track Approach Towards Automatic Detection of Gray Zo...

引用

AHFE International Conference on Cross-Cultural Decision Making (CCDM)

作者： Schlachter, Jason Lautenschlager, Jennifer Lockheed Martin Adv Technol Labs Suite 250 1825 Barrett Lakes Blvd NW Kennesaw GA 30144 USA

ISBN: (纸本)9783319607474;9783319607467

Gray Zone Operations (GZ Ops) are increasingly being used by countries to achieve national security objectives that have historically only been achieved through armed conflict. GZ Ops employ actions that are diverse, covert, and adversarial but fall below the threshold required to trigger an aggressive response from major powers. We aim to help analysts with early identification of Gray Zone Operations by automatically detecting taxonomy-backed Gray Zone activities in news stories and text. We map our proven geo-political event extraction and data processing capabilities previously developed under the World-Wide Integrated Crisis Early Warning System (W-ICEWS) program to Gray Zone activities and apply these mappings to text from a corpus of Russian news stories from 2015 to 2016 to generate a set of sentences labeled with specific Gray Zone activities. We evaluate the validity of these matches to determine if the sentences and stories accurately reflect their label.

关键词： Gray zone NLP Text analytics event coding Data mining Taxonomy

来源：评论

学校读者我要写书评

暂无评论

Evaluating Automatic Learning of Structure for event Extraction

引用

International Conference on Cross-Cultural Decision Making (CCDM)

作者： Schlachter, Jason Van Brackle, David Reynoso, Luis Asencios Starz, James Chambers, Nathanael Lockheed Martin Adv Technol Labs Informat Lab Kennesaw GA 30144 USA US Naval Acad Dept Comp Sci Annapolis MD 21402 USA

ISBN: (纸本)9783319416366;9783319416359

Analysts engaged in monitoring and forecasting benefit from the structured representations of domain knowledge and societal events that allow for the use of advanced analytics and predictive data models over large amounts of temporally extended data. However, extracting structured data from unstructured data typically requires the development of domain specific software which is costly, takes months to years to create, and cannot adapt to changing domains. In this paper we consider the operational usefulness of an approach pioneered by Chambers and Jurafsky(Template-based information extraction without the templates, 2011, [1]) that performs automatic learning of structured domain knowledge in the form of event templates from unstructured text that are used to automatically extract structured events from text. We generalize this approach and apply it to operationally relevant corpora from Brazil, Mexico, Ukraine, and Pakistan that focus on societal protests and providing aid. We discover that we are able to generate compelling event templates that correspond to event types described by Conflict and Mediation event Observations (CAMEO) codes (Retrieved from Computational event Data System, 2014, [2]) which are used to label event types by existing state of the art systems. Additionally, we are able to learn event templates that capture more nuance than the CAMEO codes represent, as well as entirely new and interesting event types. To automate our experimentation, we describe novel automated metrics that allow us to batch run multiple experiments while getting automated feedback on the quality of results from each run. These metrics indicate significant overlap between the events we extract and those extracted by existing systems.

关键词： NLP Machine learning Artificial intelligence Data analytics Text analytics Forecasting event extracting event coding

来源：评论

学校读者我要写书评

暂无评论

Extracting Meaningful Entities from Human-Generated Tactical Reports

Extracting Meaningful Entities from Human-Generated Tactical...

引用

Conference on Engineering Cyber Physical Systems - Machine Learning, Data Analytics and Smart Systems Architecting

作者： Guo, Jinhong K. Van Brackle, David LoFaso, Nicolas Hofmann, Martin O. Lockheed Martin Adv Technol Labs 3 Execut CampusSuite 600 Cherry Hill NJ 08002 USA

Military intelligence analysts use automated tools to exploit physics-based sensor data to construct a spatio-temporal picture of adversary entities, networks, and behaviors on the battlefield. Traditionally, tools did not exploit human generated, textual reports, leaving analysts to manually map dots on the map into meaningful entities using background knowledge about adversary equipment, organization, and activity. Current off-the-shelf text extraction techniques underperform on tactical reports due to unique characteristics of the text. Tactical reports typically feature short sentences with simple grammar, but also tend to include jargon and abbreviations, do not follow grammatical rules, and are likely to have spelling errors. Likewise, named entity recognizers have low recall, because few of the names in reports appear in standard dictionaries. We have developed an entity extraction capability tailored to these challenges, and to the specific needs of analysts, as part of a comprehensive exploitation and fusion system. With fewer cues from syntax, our approach uses semantic constraints to disambiguate syntactic patterns, implemented by a hybrid system that post-processes the output from a standard Natural Language Processing (NLP) engine with our custom semantic pattern analysis. Additional functionality extracts military time and location formats - essential elements that enable downstream fusion of extracted entities with sensor information resulting in a compact and meaningful representation of the battlefield situation. (C) 2015 The Authors. Published by Elsevier B.V.

关键词： Entity Extraction event coding NLP Machine Learning Information Fusion

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：