Search results for: convolution neural network- long short term memory network (CNN-LSTM)

Authors: Micheal Olaolu Arowolo, Muhammad Azam, Fei He, Mihail Popescu, Dong Xu

Abstract:

As the quantity of biological articles rises, so does the number of biological route figures. Each route figure shows gene names and relationships. Annotating pathway diagrams manually is time-consuming. Advanced image understanding models could speed up curation, but they must be more precise. There is rich information in biological pathway figures. The first step to performing image understanding of these figures is to recognize gene names automatically. Classical optical character recognition methods have been employed for gene name recognition, but they are not optimized for literature mining data. This study devised a method to recognize an image bounding box of gene name as a photo using deep Siamese neural network models to outperform the existing methods using ResNet, DenseNet and Inception architectures, the results obtained about 84% accuracy.

Keywords: biological pathway, gene identification, object detection, Siamese network

Procedia PDF Downloads 264

Authors: Hosuk Shin, Young-Hyun Seo, Eunhak Lee, Seung-Young Kho

Abstract:

Currently, in Seoul, users have the privilege to avoid riding crowded buses with the installation of Bus Information System (BIS). BIS has three levels of on-board bus ridership level information (spacious, normal, and crowded). However, there are flaws in the system due to it being real time which could provide incomplete information to the user. For example, a bus comes to the station, and on the BIS it shows that the bus is crowded, but on the stop that the user is waiting many people get off, which would mean that this station the information should show as normal or spacious. To fix this problem, this study predicts the bus ridership level using smart card data to provide more accurate information about the passenger ridership level on the bus. An Artificial Neural Network (ANN) is an interconnected group of nodes, that was created based on the human brain. Forecasting has been one of the major applications of ANN due to the data-driven self-adaptive methods of the algorithm itself. According to the results, the ANN algorithm was stable and robust with somewhat small error ratio, so the results were rational and reasonable.

Keywords: smartcard data, ANN, bus, ridership

Procedia PDF Downloads 152

Authors: Changqiao Wu, Guoqing Ding, Xin Chen

Abstract:

In this paper, ways of modeling dynamic measurement systems are discussed. Specially, for linear system with single-input single-output, it could be modeled with shallow neural network. Then, gradient based optimization algorithms are used for searching the proper coefficients. Besides, method with normal equation and second order gradient descent are proposed to accelerate the modeling process, and ways of better gradient estimation are discussed. It shows that the mathematical essence of the learning objective is maximum likelihood with noises under Gaussian distribution. For conventional gradient descent, the mini-batch learning and gradient with momentum contribute to faster convergence and enhance model ability. Lastly, experimental results proved the effectiveness of second order gradient descent algorithm, and indicated that optimization with normal equation was the most suitable for linear dynamic models.

Keywords: dynamic system modeling, neural network, normal equation, second order gradient descent

Procedia PDF Downloads 111

Authors: Hao Yan, Xiaobing Zhang

Abstract:

The working mechanism of miniaturized solid rocket motors (SRMs) is not yet fully understood. It is imperative to explore its unique features. However, there are many disadvantages to using common multi-objective evolutionary algorithms (MOEAs) in predicting the parameters of the miniaturized SRM during its conceptual design phase. Initially, the design variables and objectives are constrained in a lumped parameter model (LPM) of this SRM, which leads to local optima in MOEAs. In addition, MOEAs require a large number of calculations due to their population strategy. Although the calculation time for simulating an LPM just once is usually less than that of a CFD simulation, the number of function evaluations (NFEs) is usually large in MOEAs, which makes the total time cost unacceptably long. Moreover, the accuracy of the LPM is relatively low compared to that of a CFD model due to its assumptions. CFD simulations or experiments are required for comparison and verification of the optimal results obtained by MOEAs with an LPM. The conceptual design phase based on MOEAs is a lengthy process, and its results are not precise enough due to the above shortcomings. An artificial neural network (ANN) based parameter prediction is proposed as a way to reduce time costs and improve prediction accuracy. In this method, an ANN is used to build a surrogate model that is trained with a 3D numerical simulation. In design, the original LPM is replaced by a surrogate model. Each case uses the same MOEAs, in which the calculation time of the two models is compared, and their optimization results are compared with 3D simulation results. Using the surrogate model for the parameter prediction process of the miniaturized SRMs results in a significant increase in computational efficiency and an improvement in prediction accuracy. Thus, the ANN-based surrogate model does provide faster and more accurate parameter prediction for an initial design scheme. Moreover, even when the MOEAs converge to local optima, the time cost of the ANN-based surrogate model is much lower than that of the simplified physical model LPM. This means that designers can save a lot of time during code debugging and parameter tuning in a complex design process. Designers can reduce repeated calculation costs and obtain accurate optimal solutions by combining an ANN-based surrogate model with MOEAs.

Keywords: artificial neural network, solid rocket motor, multi-objective evolutionary algorithm, surrogate model

Procedia PDF Downloads 73

Authors: E. Behmanesh, J. Pannek

Abstract:

Logistics network design is known as one of the strategic decision problems. As these kinds of problems belong to the category of NP-hard problems, traditional ways are failed to find an optimal solution in short time. In this study, we attempt to involve reverse flow through an integrated design of forward/reverse supply chain network that formulated into a mixed integer linear programming. This Integrated, multi-stages model is enriched by three different delivery path which makes the problem more complex. To tackle with such an NP-hard problem a revised random path direct encoding method based memetic algorithm is considered as the solution methodology. Each algorithm has some parameters that need to be investigate to reveal the best performance. In this regard, Taguchi method is adapted to identify the optimum operating condition of the proposed memetic algorithm to improve the results. In this study, four factors namely, population size, crossover rate, local search iteration and a number of iteration are considered. Analyzing the parameters and improvement in results are the outlook of this research.

Keywords: integrated logistics network, flexible path, memetic algorithm, Taguchi method

Procedia PDF Downloads 175

Authors: Hicham Benamirouche, Oum Elkheir Moussi

Abstract:

The purpose of the study is to examine the dynamics of Algeria’s natural gas exports through the Autoregressive Distributed Lag (ARDL) bounds testing approach with break points. The analysis was carried out for the period from 1967 to 2015. Based on imperfect substitution specification, the ARDL approach reveals a long-run equilibrium relationship between Algeria’s Natural gas exports and their determinant factors (Algeria’s gas reserves, Domestic gas consumption, Europe’s GDP per capita, relative prices, the European gas production and the market share of competitors). All the long-run elasticities estimated are statistically significant with a large impact of domestic factors, which constitute the supply constraints. In short term, the elasticities are statistically significant, and almost comparable to those of the long term. Furthermore, the speed of adjustment towards long-run equilibrium is less than one year because of the little flexibility of the long term export contracts. Two break points have been estimated when we employ the domestic gas consumption as a break variable; 1984 and 2010, which reflect the arbitration policy between the domestic gas market and gas exports.

Keywords: natural gas exports, elasticity, ARDL bounds testing, break points, Algeria

Procedia PDF Downloads 182

Authors: Qin Long, Li Xiaoge

Abstract:

The Knowledge Graphs have now become a new form of knowledge representation. However, there is no consensus in regard to a plausible and definition of entities and relationships in the domain-specific knowledge graph. Further, in conjunction with several limitations and deficiencies, various domain-specific entities and relationships recognition approaches are far from perfect. Specifically, named entity recognition in Chinese domain is a critical task for the natural language process applications. However, a bottleneck problem with Chinese named entity recognition in new domains is the lack of annotated data. To address this challenge, a domain distant supervised named entity recognition framework is proposed. The framework is divided into two stages: first, the distant supervised corpus is generated based on the entity linking model of graph attention neural network; secondly, the generated corpus is trained as the input of the distant supervised named entity recognition model to train to obtain named entities. The link model is verified in the ccks2019 entity link corpus, and the F1 value is 2% higher than that of the benchmark method. The re-pre-trained BERT language model is added to the benchmark method, and the results show that it is more suitable for distant supervised named entity recognition tasks. Finally, it is applied in the computer field, and the results show that this framework can obtain domain named entities.

Keywords: distant named entity recognition, entity linking, knowledge graph, graph attention neural network

Procedia PDF Downloads 80

Authors: Bashir Abdu Muzakkari, Aisha Umar Sulaiman, Mohamed Afendee Muhamad, Sanah Abdullahi Muaz

Abstract:

The field of the Internet of Things (IoT) is rapidly expanding and has the potential to completely change how we work, live, and interact with the world. The Internet of Things (IoT) is the term used to describe a network of networked physical objects, including machinery, vehicles, and buildings, which are equipped with electronics, software, sensors, and network connectivity. This review paper aims to provide a comprehensive overview of the current state of IoT, including its definition, key components, development history, and current applications. The paper will also discuss the challenges and opportunities presented by IoT, as well as its potential impact on various industries, such as healthcare, agriculture, and transportation. In addition, this paper will highlight the ethical and security concerns associated with IoT and the need for effective solutions to address these challenges. The paper concludes by highlighting the prospects of IoT and the directions for future research in this field.

Keywords: internet of things, IoT, sensors, network

Procedia PDF Downloads 156

Authors: Hosam Alzahrani, Martin Mackey, Emmanuel Stamatakis, Marina B. Pinheiro, Manuela Wicks, Debra Shirley

Abstract:

Objective: To investigate the effectiveness of incidental (non-structured) physical activity interventions compared with other commonly prescribed interventions for the management of people with low back pain (LBP). Methods: We performed a systematic review with meta-analyses of eligible randomized controlled trials obtained by searching Medline, Scopus, CINAHL, EMBASE, and CENTRAL. This review considered trials investigating the effect of incidental physical activity interventions compared to other interventions in people aged 18 years or over, diagnosed with non-specific LBP. Analyses were conducted separately for short-term (≤3 months), intermediate-term (> 3 and < 12 months), and long-term (≥ 12 months), for each outcome. The analyses were conducted using the weighted mean difference (WMD). The overall quality of evidence was assessed using the GRADE system. Meta-analyses were only performed for pain and disability outcomes as there was insufficient data on the other outcomes. Results: For pain, the pooled results did not show any significant effects between the incidental physical activity intervention and other interventions at any time point. For disability, incidental physical activity was not statistically more effective than other interventions at short-term; however, the pooled results favored incidental physical activity at intermediate-term (WMD= -6.05, 95% CI: -10.39 to -1.71, p=0.006) and long-term (WMD= -6.40 95% CI: -11.68 to -1.12, p=0.02) follow-ups among participants with chronic LBP. The overall quality of evidence was rated “moderate quality” based on the GRADE system. Conclusion: The incidental physical activity intervention provided intermediate and long disability relief for people with chronic LBP, although this improvement was small and not likely to be clinically important.

Keywords: physical activity, incidental, low back pain, systematic review, meta-analysis

Procedia PDF Downloads 140

Authors: Takashi Yamaguchi, Ichio Asanuma, Jong G. Park, Kenneth J. Mackin, John Mittleman

Abstract:

In this paper, vessel detection using the artificial neural network is proposed in order to automatically construct the vessel detection model from the satellite imagery of day/night band (DNB) in visible infrared in the products of Imaging Radiometer Suite (VIIRS) on Suomi National Polar-orbiting Partnership (Suomi-NPP).The goal of our research is the establishment of vessel detection method using the satellite imagery of DNB in order to monitor the change of vessel activity over the wide region. The temporal vessel monitoring is very important to detect the events and understand the circumstances within the maritime environment. For the vessel locating and detection techniques, Automatic Identification System (AIS) and remote sensing using Synthetic aperture radar (SAR) imagery have been researched. However, each data has some lack of information due to uncertain operation or limitation of continuous observation. Therefore, the fusion of effective data and methods is important to monitor the maritime environment for the future. DNB is one of the effective data to detect the small vessels such as fishery ships that is difficult to observe in AIS. DNB is the satellite sensor data of VIIRS on Suomi-NPP. In contrast to SAR images, DNB images are moderate resolution and gave influence to the cloud but can observe the same regions in each day. DNB sensor can observe the lights produced from various artifact such as vehicles and buildings in the night and can detect the small vessels from the fishing light on the open water. However, the modeling of vessel detection using DNB is very difficult since complex atmosphere and lunar condition should be considered due to the strong influence of lunar reflection from cloud on DNB. Therefore, artificial neural network was applied to learn the vessel detection model. For the feature of vessel detection, Brightness Temperature at the 3.7 μm (BT3.7) was additionally used because BT3.7 can be used for the parameter of atmospheric conditions.

Keywords: artificial neural network, day/night band, remote sensing, Suomi National Polar-orbiting Partnership, vessel detection, Visible Infrared Imaging Radiometer Suite

Procedia PDF Downloads 222

Authors: Mamadou Dione, Eric Matzner-lober, Philippe Alexandre

Abstract:

The Energy Transition Act defined by the French State has precise implications on Renewable Energies, in particular on its remuneration mechanism. Until then, a purchase obligation contract permitted the sale of wind-generated electricity at a fixed rate. Tomorrow, it will be necessary to sell this electricity on the Market (at variable rates) before obtaining additional compensation intended to reduce the risk. This sale on the market requires to announce in advance (about 48 hours before) the production that will be delivered on the network, so to be able to predict (in the short term) this production. The fundamental problem remains the variability of the Wind accentuated by the geographical situation. The objective of the project is to provide, every day, short-term forecasts (48-hour horizon) of wind production using weather data. The predictions of the GFS model and those of the ECMWF model are used as explanatory variables. The variable to be predicted is the production of a wind farm. We do two approaches: a direct approach that predicts wind generation directly from weather data, and an integrated approach that estimâtes wind from weather data and converts it into wind power by power curves. We used machine learning techniques to predict this production. The models tested are random forests, CART + Bagging, CART + Boosting, SVM (Support Vector Machine). The application is made on a wind farm of 22MW (11 wind turbines) of the Compagnie du Vent (that became Engie Green France). Our results are very conclusive compared to the literature.

Keywords: forecast aggregation, machine learning, spatio-temporal dynamics modeling, wind power forcast

Procedia PDF Downloads 198

Authors: Jonathan Sohn

Abstract:

Deep neural networks (DNNs) have shown state-of-the-art performance for many applications, including computer vision, natural language processing, and speech recognition. Recently, adversarial attacks have been studied in the context of deep neural networks, which aim to alter the results of deep neural networks by modifying the inputs slightly. For example, an adversarial attack on a DNN used for object detection can cause the DNN to miss certain objects. As a result, the reliability of DNNs is undermined by their lack of robustness against adversarial attacks, raising concerns about their use in safety-critical applications such as autonomous driving. In this paper, we focus on studying the adversarial attacks and defenses on DNNs for image classification. There are two types of adversarial attacks studied which are fast gradient sign method (FGSM) attack and projected gradient descent (PGD) attack. A DNN forms decision boundaries that separate the input images into different categories. The adversarial attack slightly alters the image to move over the decision boundary, causing the DNN to misclassify the image. FGSM attack obtains the gradient with respect to the image and updates the image once based on the gradients to cross the decision boundary. PGD attack, instead of taking one big step, repeatedly modifies the input image with multiple small steps. There is also another type of attack called the target attack. This adversarial attack is designed to make the machine classify an image to a class chosen by the attacker. We can defend against adversarial attacks by incorporating adversarial examples in training. Specifically, instead of training the neural network with clean examples, we can explicitly let the neural network learn from the adversarial examples. In our experiments, the digit recognition accuracy on the MNIST dataset drops from 97.81% to 39.50% and 34.01% when the DNN is attacked by FGSM and PGD attacks, respectively. If we utilize FGSM training as a defense method, the classification accuracy greatly improves from 39.50% to 92.31% for FGSM attacks and from 34.01% to 75.63% for PGD attacks. To further improve the classification accuracy under adversarial attacks, we can also use a stronger PGD training method. PGD training improves the accuracy by 2.7% under FGSM attacks and 18.4% under PGD attacks over FGSM training. It is worth mentioning that both FGSM and PGD training do not affect the accuracy of clean images. In summary, we find that PGD attacks can greatly degrade the performance of DNNs, and PGD training is a very effective way to defend against such attacks. PGD attacks and defence are overall significantly more effective than FGSM methods.

Keywords: deep neural network, adversarial attack, adversarial defense, adversarial machine learning

Procedia PDF Downloads 177

Authors: Omar M. Aldenali

Abstract:

Planning is one of the most important phases of the management science and network planning, which represents the project activities relationship. Critical path is one of the project management techniques used to plan and control the execution of a project activities. The objective of this paper is to implement a fuzzy logic approach to arrange network planning on construction projects. This method is used to finding out critical path in the fuzzy construction project network. The trapezoidal fuzzy numbers are used to represent the activity construction project times. A numerical example that represents a house construction project is introduced. The critical path method is implemented on the fuzzy construction network activities, and the results showed that this method significantly affects the completion time of the construction projects.

Keywords: construction project, critical path, fuzzy network project, planning

Procedia PDF Downloads 126

Authors: Hafiz Gulfam Ahmad, Chuangdong Li, Zeeshan Ahmad

Abstract:

In this paper, we proposed a parallel IDS and honeypot based approach to detect and analyze the unknown and known attack taxonomy for improving the IDS performance and protecting the network from intruders. The main theme of our approach is to record and analyze the intruder activities by using both the low and high interaction honeypots. Our architecture aims to achieve the required goals by combing signature based IDS, honeypots and generate the new signatures. The paper describes the basic component, design and implementation of this approach and also demonstrates the effectiveness of this approach reducing the probability of network attacks.

Keywords: network security, intrusion detection, honeypot, snort, nmap

Procedia PDF Downloads 537

Authors: Douglas Rossi Ramos

Abstract:

The current digital conjuncture, called by some authors as 'Internet of Things' (IoT), 'Web 2.0' or even 'Web 3.0', consists of a network that encompasses any communication of objects and entities, such as data, information, technologies, and people. At this juncture, especially characterized by an "object socialization," communication can no longer be represented as a simple informational flow of messages from a sender, crossing a channel or medium, reaching a receiver. The idea of communication must, therefore, be thought of more broadly in which it is possible to analyze the process communicative from interactions between humans and nonhumans. To think about this complexity, a communicative process that encompasses both humans and other beings or entities communicating (objects and things), it is necessary to constitute a new epistemology of communication to rethink concepts and notions commonly attributed to humans such as 'memory.' This research aims to contribute to this epistemological constitution from the discussion about the notion of memory according to the complex ontology of Henri Bergson. Among the results (the notion of memory in Bergson presents itself as a conceptual tool for the analysis of posthumanism and the anthropomorphic conjuncture of the new advent of digital), there was the need to think about an ontological memory, analyzed as a being itself (being itself of memory), as a strategy for understanding the forms of interaction and communication that constitute the new digital conjuncture, in which communicating beings or entities tend to interact with each other. Rethinking the idea of communication beyond the dimension of transmission in informative sequences paves the way for an ecological perspective of the digital dwelling condition.

Keywords: communication, digital, Henri Bergson, memory

Procedia PDF Downloads 142

Authors: Alexander Vandesompele, Joni Dambre

Abstract:

The diversity in dendritic arborization, as first illustrated by Santiago Ramon y Cajal, has always suggested a role for dendrites in the functionality of neurons. In the past decades, thanks to new recording techniques and optical stimulation methods, it has become clear that dendrites are not merely passive electrical components. They are observed to integrate inputs in a non-linear fashion and actively participate in computations. Regardless, in simulations of neural networks dendritic structure and functionality are often overlooked. Especially in a machine learning context, when designing artificial neural networks, point neuron models such as the leaky-integrate-and-fire (LIF) model are dominant. These models mimic the integration of inputs at the neuron soma, and ignore the existence of dendrites. In this work, the LIF point neuron model is extended with a simple form of dendritic computation. This gives the LIF neuron increased capacity to discriminate spatiotemporal input sequences, a dendritic functionality as observed in another study. Simulations of the spiking neurons are performed using the Bindsnet framework. In the common LIF model, incoming synapses are independent. Here, we introduce a dependency between incoming synapses such that the post-synaptic impact of a spike is not only determined by the weight of the synapse, but also by the activity of other synapses. This is a form of short term plasticity where synapses are potentiated or depressed by the preceding activity of neighbouring synapses. This is a straightforward way to prevent inputs from simply summing linearly at the soma. To implement this, each pair of synapses on a neuron is assigned a variable,representing the synaptic relation. This variable determines the magnitude ofthe short term plasticity. These variables can be chosen randomly or, more interestingly, can be learned using a form of Hebbian learning. We use Spike-Time-Dependent-Plasticity (STDP), commonly used to learn synaptic strength magnitudes. If all neurons in a layer receive the same input, they tend to learn the same through STDP. Adding inhibitory connections between the neurons creates a winner-take-all (WTA) network. This causes the different neurons to learn different input sequences. To illustrate the impact of the proposed dendritic mechanism, even without learning, we attach five input neurons to two output neurons. One output neuron isa regular LIF neuron, the other output neuron is a LIF neuron with dendritic relationships. Then, the five input neurons are allowed to fire in a particular order. The membrane potentials are reset and subsequently the five input neurons are fired in the reversed order. As the regular LIF neuron linearly integrates its inputs at the soma, the membrane potential response to both sequences is similar in magnitude. In the other output neuron, due to the dendritic mechanism, the membrane potential response is different for both sequences. Hence, the dendritic mechanism improves the neuron’s capacity for discriminating spa-tiotemporal sequences. Dendritic computations improve LIF neurons even if the relationships between synapses are established randomly. Ideally however, a learning rule is used to improve the dendritic relationships based on input data. It is possible to learn synaptic strength with STDP, to make a neuron more sensitive to its input. Similarly, it is possible to learn dendritic relationships with STDP, to make the neuron more sensitive to spatiotemporal input sequences. Feeding structured data to a WTA network with dendritic computation leads to a significantly higher number of discriminated input patterns. Without the dendritic computation, output neurons are less specific and may, for instance, be activated by a sequence in reverse order.

Keywords: dendritic computation, spiking neural networks, point neuron model

Procedia PDF Downloads 115

Authors: Apoorva Vinod, Archana Mathur, Snehanshu Saha

Abstract:

Though there exists a plethora of Activation Functions (AFs) used in single and multiple hidden layer Neural Networks (NN), their behavior always raised curiosity, whether used in combination or singly. The popular AFs –Sigmoid, ReLU, and Tanh–have performed prominently well for shallow and deep architectures. Most of the time, AFs are used singly in multi-layered NN, and, to the best of our knowledge, their performance is never studied and analyzed deeply when used in combination. In this manuscript, we experiment with multi-layered NN architecture (both on shallow and deep architectures; Convolutional NN and VGG16) and investigate how well the network responds to using two different AFs (Sigmoid-Tanh, Tanh-ReLU, ReLU-Sigmoid) used alternately against a traditional, single (Sigmoid-Sigmoid, Tanh-Tanh, ReLUReLU) combination. Our results show that using two different AFs, the network achieves better accuracy, substantially lower loss, and faster convergence on 4 computer vision (CV) and 15 Non-CV (NCV) datasets. When using different AFs, not only was the accuracy greater by 6-7%, but we also accomplished convergence twice as fast. We present a case study to investigate the probability of networks suffering vanishing and exploding gradients when using two different AFs. Additionally, we theoretically showed that a composition of two or more AFs satisfies Universal Approximation Theorem (UAT).

Keywords: activation function, universal approximation function, neural networks, convergence

Procedia PDF Downloads 140

Authors: Zhaopeng Zhu, Xianzhi Song, Gensheng Li, Shuo Zhu, Shiming Duan, Xuezhe Yao

Abstract:

Accurate calculation of wellbore pressure is of great significance to prevent wellbore risk during drilling. The traditional mechanism model needs a lot of iterative solving procedures in the calculation process, which reduces the calculation efficiency and is difficult to meet the demand of dynamic control of wellbore pressure. In recent years, many scholars have introduced artificial intelligence algorithms into wellbore pressure calculation, which significantly improves the calculation efficiency and accuracy of wellbore pressure. However, due to the ‘black box’ property of intelligent algorithm, the existing intelligent calculation model of wellbore pressure is difficult to play a role outside the scope of training data and overreacts to data noise, often resulting in abnormal calculation results. In this study, the multi-phase flow mechanism is embedded into the objective function of the neural network model as a constraint condition, and an intelligent prediction model of wellbore pressure under the constraint condition is established based on more than 400,000 sets of pressure measurement while drilling (MPD) data. The constraint of the multi-phase flow mechanism makes the prediction results of the neural network model more consistent with the distribution law of wellbore pressure, which overcomes the black-box attribute of the neural network model to some extent. The main performance is that the accuracy of the independent test data set is further improved, and the abnormal calculation values basically disappear. This method is a prediction method driven by MPD data and multi-phase flow mechanism, and it is the main way to predict wellbore pressure accurately and efficiently in the future.

Keywords: multiphase flow mechanism, pressure while drilling data, wellbore pressure, mechanism constraints, combined drive

Procedia PDF Downloads 162

Authors: Paola Causin, Andrea Aspri, Alessandro Benfenati

Abstract:

Diffuse Optical Tomography (DOT) is an emergent medical imaging technique which employs NIR light to estimate the spatial distribution of optical coefficients in biological tissues for diagnostic purposes, in a noninvasive and non-ionizing manner. DOT reconstruction is a severely ill-conditioned problem due to prevalent scattering of light in the tissue. In this contribution, we present our research in adopting hybrid knowledgedriven/data-driven approaches which exploit the existence of well assessed physical models and build upon them neural networks integrating the availability of data. Namely, since in this context regularization procedures are mandatory to obtain a reasonable reconstruction [1], we explore the use of neural networks as tools to include prior information on the solution. 2. Materials and Methods The idea underlying our approach is to leverage neural networks to solve PDE-constrained inverse problems of the form 𝒒 ∗ = 𝒂𝒓𝒈 𝒎𝒊𝒏𝒒 𝐃(𝒚, 𝒚̃), (1) where D is a loss function which typically contains a discrepancy measure (or data fidelity) term plus other possible ad-hoc designed terms enforcing specific constraints. In the context of inverse problems like (1), one seeks the optimal set of physical parameters q, given the set of observations y. Moreover, 𝑦̃ is the computable approximation of y, which may be as well obtained from a neural network but also in a classic way via the resolution of a PDE with given input coefficients (forward problem, Fig.1 box ). Due to the severe ill conditioning of the reconstruction problem, we adopt a two-fold approach: i) we restrict the solutions (optical coefficients) to lie in a lower-dimensional subspace generated by auto-decoder type networks. This procedure forms priors of the solution (Fig.1 box ); ii) we use regularization procedures of type 𝒒̂ ∗ = 𝒂𝒓𝒈𝒎𝒊𝒏𝒒 𝐃(𝒚, 𝒚̃)+ 𝑹(𝒒), where 𝑹(𝒒) is a regularization functional depending on regularization parameters which can be fixed a-priori or learned via a neural network in a data-driven modality. To further improve the generalizability of the proposed framework, we also infuse physics knowledge via soft penalty constraints (Fig.1 box ) in the overall optimization procedure (Fig.1 box ). 3. Discussion and Conclusion DOT reconstruction is severely hindered by ill-conditioning. The combined use of data-driven and knowledgedriven elements is beneficial and allows to obtain improved results, especially with a restricted dataset and in presence of variable sources of noise.

Keywords: inverse problem in tomography, deep learning, diffuse optical tomography, regularization

Procedia PDF Downloads 61

Authors: Veerpaul Maan, Gaurav Mishra

Abstract:

The evolution of omnichannel has revolutionized the supply chain of the organizations by enhancing customer shopping experience. For these organizations need to develop well-integrated multiple distribution channels to leverage the benefits of omnichannel. To adopt an omnichannel system in the supply chain has resulted in structuring and reconfiguring the practices of the traditional supply chain distribution network. In this paper a multiple distribution supply chain network (MDSCN) have been proposed which integrates online giants with a local retailers distribution network in uncertain environment followed by sustainability. To incorporate sustainability, an additional objective function is added to reduce the carbon content through minimizing the travel distance of the product. Through this proposed model, customers are free to access product and services as per their choice of channels which increases their convenience, reach and satisfaction. Further, a numerical illustration is being shown along with interpretation of results to validate the proposed model.

Keywords: sustainable supply chain network, omnichannel, multiple distribution supply chain network, integrate multiple distribution channels

Procedia PDF Downloads 208

Authors: Smail Boussaadi, Hassina Aliane

Abstract:

With the rapid growth of scientific literature, it is becoming increasingly challenging for researchers to keep up with the latest findings in their fields. Academic, professional networks play an essential role in connecting researchers and disseminating knowledge. To improve the user experience within these networks, we need effective article recommendation systems that provide personalized content.Current recommendation systems often rely on collaborative filtering or content-based techniques. However, these methods have limitations, such as the cold start problem and difficulty in capturing semantic relationships between articles. To overcome these challenges, we propose a new approach that combines BERTopic (Bidirectional Encoder Representations from Transformers), a state-of-the-art topic modeling technique, with community detection algorithms in a academic, professional network. Experiences confirm our performance expectations by showing good relevance and objectivity in the results.

Keywords: scientific articles, community detection, academic social network, recommender systems, neural topic model

Procedia PDF Downloads 80

Authors: Brenden M. Shutt, Lubjana Beshaj, Paul L. Goethals, Ambrose Kam

Abstract:

Communication security is of particular interest to military data networks. A relatively novel approach to network security is blockchain, a cryptographically secured distribution ledger with a decentralized consensus mechanism for data transaction processing. Recent advances in blockchain technology have proposed new techniques for both data validation and trust management, as well as different frameworks for managing dataflow. The purpose of this work is to test the feasibility of different blockchain architectures as applied to military command and control networks. Various architectures are tested through discrete-event simulation and the feasibility is determined based upon a blockchain design’s ability to maintain long-term stable performance at industry standards of throughput, network latency, and security. This work proposes a consortium blockchain architecture with a computationally inexpensive consensus mechanism, one that leverages a Proof-of-Identity (PoI) concept and a reputation management mechanism.

Keywords: blockchain, consensus mechanism, discrete-event simulation, fog computing

Procedia PDF Downloads 122

Authors: Sipu Guo, Silin Huang

Abstract:

Background: Most studies investigate the link between executive function and allostatic load (AL) among adults aged 18 years and older. Studies differed regarding the specific biological indicators studied and executive functions accounted for. Specific executive functions may be differentially related to allostatic load. We investigated the comorbidities of executive functions and allostatic load via network analysis. Methods: We included 603 adolescents (49.84% girls; Mean age = 12.38, SD age = 1.79) from junior high school in rural China. Eight biological markers at T1 and four executive function tasks at T2 were used to evaluate networks. Network analysis was used to determine the network structure, core symptoms, and bridge symptoms in the AL-executive function network among rural adolescents. Results: The executive functions were related to 6 AL biological markers, not to cortisol and epinephrine. The most influential symptoms were inhibition control, cognitive flexibility, processing speed, and systolic blood pressure (SBP). SBP, dehydroepiandrosterone, and processing speed were the bridges through which AL was related to executive functions. dehydroepiandrosterone strongly predicted processing speed. The SBP was the biggest influencer in the entire network. Conclusions: We found evidence for differential relations between markers and executive functions. SBP was a driver in the network; dehydroepiandrosterone showed strong relations with executive function.

Keywords: allostatic load, executive function, network analysis, rural adolescent

Procedia PDF Downloads 37

Authors: Vaishnavi Shivde, Shrishti Sinha, Trapti Mishra

Abstract:

Traffic light detection from a moving vehicle is an important technology both for driver safety assistance functions as well as for autonomous driving in the city. This paper proposed a deep-learning-based traffic light recognition method that consists of a pixel-wise image segmentation technique and a fully convolutional network i.e., UNET architecture. This paper has used a method for detecting the position and recognizing the state of the traffic lights in video sequences is presented and evaluated using Traffic Light Dataset which contains masked traffic light image data. The first stage is the detection, which is accomplished through image processing (image segmentation) techniques such as image cropping, color transformation, segmentation of possible traffic lights. The second stage is the recognition, which means identifying the color of the traffic light or knowing the state of traffic light which is achieved by using a Convolutional Neural Network (UNET architecture).

Keywords: traffic light detection, image segmentation, machine learning, classification, convolutional neural networks

Procedia PDF Downloads 153

Authors: Justin Reverdi, Sixin Zhang, Serge Gratton, Said Aoues, Thomas Pellegrini

Abstract:

In Vapor Cycle Systems, the flow sensor plays a key role in different monitoring and control purposes. However, physical sensors can be expensive, inaccurate, heavy, cumbersome, or highly sensitive to vibrations, which is especially problematic when embedded into an aircraft. The conception of a virtual sensor based on other standard sensors is a good alternative. In this paper, a data-driven model using a Convolutional Neural Network is proposed to estimate the flow of the compressor. To fit the model to our dataset, we tested different loss functions. We show in our application that a Dynamic Time Warping based loss function called DILATE leads to better dynamical performance than the vanilla mean squared error (MSE) loss function. DILATE allows choosing a trade-off between static and dynamic performance.

Keywords: deep learning, dynamic time warping, vapor cycle system, virtual sensor

Procedia PDF Downloads 131

Authors: Rashmi Tamhankar, Madhav Bapat

Abstract:

In many projects, we have to use human judgment for determining the duration of the activities which may vary from person to person. Hence, there is vagueness about the time duration for activities in network planning. Fuzzy sets can handle such vague or imprecise concepts and has an application to such network. The vague activity times can be represented by triangular fuzzy numbers. In this paper, a general network with fuzzy activity times is considered and conditions for the critical path are obtained also we compute total float time of each activity. Several numerical examples are discussed.

Keywords: PERT, CPM, triangular fuzzy numbers, fuzzy activity times

Procedia PDF Downloads 456

Authors: Man Fung Ho, Lap So, Jiaqi Zhang, Yuheng Zhao, Huiyang Lu, Tat Shing Choi, K. Y. Michael Wong

Abstract:

Nowadays, a tremendous amount of data is available in the transportation system, enabling the development of various machine learning approaches to make short-term latency predictions. A natural question is then the choice of relevant information to enable accurate predictions. Using traffic data collected from the Taiwan Freeway System, we consider the prediction of short-term latency of a freeway segment with a length of 17 km covering 5 measurement points, each collecting vehicle-by-vehicle data through the electronic toll collection system. The processed data include the past latencies of the freeway segment with different time lags, the traffic conditions of the individual segments (the accumulations, the traffic fluxes, the entrance and exit rates), the total accumulations, and the weekday latency profiles obtained by Gaussian process regression of past data. We arrive at several important conclusions about how data should be refined to obtain accurate predictions, which have implications for future system-wide latency predictions. (1) We find that the prediction of median latency is much more accurate and meaningful than the prediction of average latency, as the latter is plagued by outliers. This is verified by machine-learning prediction using XGBoost that yields a 35% improvement in the mean square error of the 5-minute averaged latencies. (2) We find that the median latency of the segment 15 minutes ago is a very good baseline for performance comparison, and we have evidence that further improvement is achieved by machine learning approaches such as XGBoost and Long Short-Term Memory (LSTM). (3) By analyzing the feature importance score in XGBoost and calculating the mutual information between the inputs and the latencies to be predicted, we identify a sequence of inputs ranked in importance. It confirms that the past latencies are most informative of the predicted latencies, followed by the total accumulation, whereas inputs such as the entrance and exit rates are uninformative. It also confirms that the inputs are much less informative of the average latencies than the median latencies. (4) For predicting the latencies of segments composed of two or three sub-segments, summing up the predicted latencies of each sub-segment is more accurate than the one-step prediction of the whole segment, especially with the latency prediction of the downstream sub-segments trained to anticipate latencies several minutes ahead. The duration of the anticipation time is an increasing function of the traveling time of the upstream segment. The above findings have important implications to predicting the full set of latencies among the various locations in the freeway system.

Keywords: data refinement, machine learning, mutual information, short-term latency prediction

Procedia PDF Downloads 158

Authors: Muhammad Kamran Asif

Abstract:

MPLS stands for Multi-Protocol Label Switching. It is the technology which replaces ATM (Asynchronous Transfer Mode) and frame relay. In this paper, we have designed a full fledge small scale MPLS based service provider network core network model, which provides communication services (e.g. voice, video and data) to the customer more efficiently using label switching technique. Using MPLS VPN provides security to the customers which are either on LAN or WAN. It protects its single customer sites from being attacked by any intruder from outside world along with the provision of concept of extension of a private network over an internet. In this paper, we tried to implement a service provider network using minimum available resources i.e. five 3800 series CISCO routers comprises of service provider core, provider edge routers and customer edge routers. The customers on the one end of the network (customer side) is capable of sending any kind of data to the customers at the other end using service provider cloud which is MPLS VPN enabled. We have also done simulation and emulation for the model using GNS3 (Graphical Network Simulator-3) and achieved the real time scenarios. We have also deployed a NMS system which monitors our service provider cloud and generates alarm in case of any intrusion or malfunctioning in the network. Moreover, we have also provided a video help desk facility between customers and service provider cloud to resolve the network issues more effectively.

Keywords: MPLS, VPN, NMS, ATM, asynchronous transfer mode

Procedia PDF Downloads 318

Authors: Juha Kettunen

Abstract:

This study analyzes the quality and the size of the strategic network of higher education institutions. The study analyses the concept of fitness for purpose in quality assurance. It also analyses the transaction costs of networking that have consequences on the number of members in the network. Empirical evidence is presented of the Consortium on Applied Research and Professional Education, which is a European strategic network of six higher education institutions. The results of the study support the argument that the number of members in the strategic network should be relatively small to provide high quality results. The practical importance is that networking has been able to promote international research and development projects. The results of this study are important for those who want to design and improve international networks in higher education.

Keywords: balanced scorecard, higher education, social networking, strategic planning

Procedia PDF Downloads 327

Authors: Mohamed Mokhtar, Mostafa F. Shaaban

Abstract:

Among the challenges facing solar photovoltaic (PV) systems in the United Arab Emirates (UAE), dust accumulation on solar panels is considered the most severe problem that faces the growth of solar power plants. The accumulation of dust on the solar panels significantly degrades output from these panels. Hence, solar PV panels have to be cleaned manually or using costly automated cleaning methods. This paper focuses on initiating cleaning actions when required to reduce maintenance costs. The cleaning actions are triggered only when the dust level exceeds a threshold value. The amount of dust accumulated on the PV panels is estimated using an artificial neural network (ANN). Experiments are conducted to collect the required data, which are used in the training of the ANN model. Then, this ANN model will be fed by the output power from solar panels, ambient temperature, and solar irradiance, and thus, it will be able to estimate the amount of dust accumulated on solar panels at these conditions. The model was tested on different case studies to confirm the accuracy of the developed model.

Keywords: machine learning, dust, PV panels, renewable energy

Procedia PDF Downloads 127