This article develops a low-cost multi-channel impedance measurement device as a teaching tool based on the electrical impedance tomography (EIT) technique. The device is applied for the teaching of "signal analy...
详细信息
As the United States Air Force moves towards autonomous labelling of FMV from ISR sensors, it has experienced unforeseen technical and legal challenges. In terms of the technical challenges, this research effort ident...
详细信息
ISBN:
(纸本)9781450365291
As the United States Air Force moves towards autonomous labelling of FMV from ISR sensors, it has experienced unforeseen technical and legal challenges. In terms of the technical challenges, this research effort identifies these obstacles and presents solutions for them with detailed step-by-step analysis of the processes, its testing and prototypes. In terms of the legal challenges, the USAF's goals of infusing artificial intelligence into autonomous labelling of FMV is also being challenged by a formidable, looming legal threat of new laws that will force the USAF to include 'humans in the loop' of its artificial intelligence andmachinelearning systems [20], [7], [15]. Again, we analyze these legal threats and present solutions to allow inclusion of a human in the loop. It is important to note that our solution to these technical and legal challenges form a two-pronged solution that yields a Bench to Battlefield, Government off-the-shelf (GOTS) autonomous FMV labelling system that will, as time goes by, learn and grow in its ISR identification abilities.
The changes in the trading market are affected by many factors, and traditional forecasting methods are more and more difficult to meet people's needs. In order to improve the accuracy of prediction, this paper pr...
详细信息
This paper presents a system for extrapolating knowledge and classification rules from existing ISR FMV and creating an ISR-Brain. As combat operations have grown to depend upon assured, live ISR support during operat...
详细信息
ISBN:
(纸本)9781450365291
This paper presents a system for extrapolating knowledge and classification rules from existing ISR FMV and creating an ISR-Brain. As combat operations have grown to depend upon assured, live ISR support during operations, US forces are presented with formidable challenges to integrate artificial intelligence (AI) capabilities with existing ISR systems. The common challenge being the variance at which advances in commercial and academic AI are deployed compared to rate of speed that innovative AI systems are developed and utilized in military domains. ISR, USAF and SOCOM need to develop a means to seamlessly integrate military and commercial state-of-the-art systems. The ISR-Brain presented will be capable of converting classifiers in existing ISR FMV to machinelearning rules for real time ISR sensor, multi-source, multi-enclave data and adaptable with ongoing research efforts with A2, SOCOM, REDO, MITRE and Project MAVEN to develop and test and ISR-Brain to enable the system to integrate with all ISR sensors and predict future Troops in Contact events (TIC) and IED events.
On the basis of the difficulty in determining the feature point in the signalprocessing of gas ultrasonic flowmeter,a variable threshold based zero-crossing detection signalprocessing method is proposed to determine...
详细信息
Sentiment analysis has recently been considered as most active research field in NLP domain. Deep learning is a growing trend of machinelearning due to its automatic learning capability with impressive results across...
详细信息
ISBN:
(纸本)9781450364027
Sentiment analysis has recently been considered as most active research field in NLP domain. Deep learning is a growing trend of machinelearning due to its automatic learning capability with impressive results across different NLP task. In this paper a model is proposed to analyze the deep sentiment representation based on CNN and LSTM (modified version of RNN) network. We aim to improve the performance of traditional machinelearning method by merging them with deep learning techniques to tackle the challenge of sentiment prediction of massive amount of unsupervised product review dataset. We make our model first learn to sentence representation with CNN. Next, the semantics of sentences are encoded with LSTM network for document representation. We conduct experiments on two review datasets based on movie review with evaluation metric 'accuracy'. The result shows that proposed model outperformed traditional machinelearning as well as baseline neural network model
The paper considers the possibilities of using neural network methods of machinelearning to diagnose the states of the human cardiovascular system and support decision-making in cardiology and cardiac surgery. The is...
详细信息
ISBN:
(数字)9781510634107
ISBN:
(纸本)9781510634107
The paper considers the possibilities of using neural network methods of machinelearning to diagnose the states of the human cardiovascular system and support decision-making in cardiology and cardiac surgery. The issues of processing and preparation of electrocardiography signals, selection of architecture and tuning of neural network parameters for automation of diagnosis are discussed. Here, the results obtained with the help of multilayer perceptrons and convolutional neural networks to assign the submitted input cardiovascular data to one of the classes of states in the selected space are examined. Based on a specialized developed software, the proprietary numerical experiments with real clinical data were carried out. Given the above results, demonstrating the applicability of the used deep learning methods and algorithms to diagnostic automation, a model of a hierarchical decision support system is proposed.
In modern sonar systems, automatic recognition of underwater targets has always been one of the key technologies in research. In recent years, classification and recognition methods based on machinelearning have been...
详细信息
ISBN:
(纸本)9781728151021
In modern sonar systems, automatic recognition of underwater targets has always been one of the key technologies in research. In recent years, classification and recognition methods based on machinelearning have been widely used in underwater acoustic field where good results have been achieved. Compared with monostatic active sonar, multi-static active sonar can simultaneously acquire the forward, lateral, and backscattering information of the target, and can obtain more accurate and stable target recognition result. Furthermore, the transmit waveform of active sonar effort the performance in complex ocean environment. As all known, the signals transmitted by cetaceans have the characteristics of strong anti-jamming ability, high positioning accuracy, et al. Accordingly, the performance of multi-static active sonar target recognition based on bionic signal is investigated in this paper. Besides, the machinelearning methods are applied to the recognition of echo signals, so that further good results and conclusions are obtained.
The background of this paper is new social trend of more public's interest in the implementation of the pledge of the local governors who were elected by citizens. In these days the election pledge for enhanced lo...
详细信息
ISBN:
(纸本)9781450366427
The background of this paper is new social trend of more public's interest in the implementation of the pledge of the local governors who were elected by citizens. In these days the election pledge for enhanced local governmental policies became more important. The objective of this paper is to suggest the model of election pledge management for local government heads based on machinelearning focused on On-Nara document system. The system is currently used by Korean governmental organizations for document processes. The methods to prove a comparative advantage of the proposed model are the comparison tests between As-Is system and To-Be system based on a few criteria such as time, efficiency and extraction rate. Through this model, local governors could present systematic goals and road map of pledges in order to get closer to citizens and local residents. In other words, this study proposes a model, so called ELM(Election pledge management for Local governors Model), for efficiently extracting necessary data from planned and implemented details of pledge projects that are prepared in the form of unstructured documents. We carried out research to prove empirically our machinelearning-based model is more efficient than current semi-manual system with some automated processes in order to manage efficiently the pledge project implementation of local governors to get the results. In conclusion, this research proved that the proposed model is more competitive than the existing models. In the 4 th industrial revolution era the new approach using machinelearning and big data will become more popular.
As infrasonic signals can through objects and propagate at a long distance, infrasound sensors are widely applied in wireless sensor networks to monitor environment events of a large area. The signal conditions are us...
详细信息
ISBN:
(纸本)9781450364027
As infrasonic signals can through objects and propagate at a long distance, infrasound sensors are widely applied in wireless sensor networks to monitor environment events of a large area. The signal conditions are usually complex and have various characteristics while monitoring the large area. Different features in both time and frequency domains should be extracted and considered. Big data increases the computation complexity, and the wrong selection of features may decreases the accuracy in event prediction. To overcome this problem, a query-based learning method is applied to select the proper features for smart edge computing in machinelearning. Experimental results show that the proposed method provides good performance when comparing with previous feature selection methods.
暂无评论