Search results for: deep mixing method

Authors: Reza Hadjiaghaie Vafaie, Sevda Givtaj

Abstract:

Automation and control of biological samples and solutions at the microscale is a major advantage for biochemistry analysis and biological diagnostics. Despite the known potential of miniaturization in biochemistry and biomedical applications, comparatively little is known about fluid automation and control at the microscale. Here, we study the electric field effect inside a fluidic channel and proper electrode structures with different patterns proposed to form forward, reversal, and rotational flows inside the channel. The simulation results confirmed that the ac electro-thermal flow is efficient for the control and automation of high-conductive solutions. In this research, the fluid pumping and mixing effects were numerically studied by solving physic-coupled electric, temperature, hydrodynamic, and concentration fields inside a microchannel. From an experimental point of view, the electrode structures are deposited on a silicon substrate and bonded to a PDMS microchannel to form a microfluidic chip. The motions of fluorescent particles in pumping and mixing modes were captured by using a CCD camera. By measuring the frequency response of the fluid and exciting the electrodes with the proper voltage, the fluid motions (including pumping and mixing effects) are observed inside the channel through the CCD camera. Based on the results, there is good agreement between the experimental and simulation studies.

Keywords: microfluidic, nano/micro actuator, AC electrothermal, Reynolds number, micropump, micromixer, microfabrication, mass transfer, biomedical applications

Procedia PDF Downloads 48

Authors: Lihong Wu, Raymond Young

Abstract:

The ability to learn is one of the most critical skills in the information age. However, many students do not have a clear understanding of what learning is, what they are learning, and why they are learning. Many students study simply to pass rather than to learn something useful for their career and their life. They have a misconception about learning and a wrong attitude towards learning. This research explores student attitudes to study in management education and explores how to intercede to lead students from shallow to deeper modes of learning.

Keywords: knowledge, surface learning, deep learning, education

Procedia PDF Downloads 471

Authors: Yong Zhang

Abstract:

Light has proven to be useful in a wide range of biomedical applications such as fluorescence imaging, photoacoustic imaging, optogenetics, photodynamic therapy, photothermal therapy, and light controlled drug/gene delivery. Taking photodynamic therapy (PDT) as an example, PDT has been proven clinically effective in early lung cancer, bladder cancer, head, and neck cancer and is the primary treatment for skin cancer as well. However, clinical use of PDT is severely constrained by the low penetration depth of visible light through thick tissue, limiting its use to target regions only a few millimeters deep. One way to enhance the range is to use invisible near-infrared (NIR) light within the optical window (700–1100nm) for biological tissues, extending the depth up to 1cm with no observable damage to the intervening tissue. We have demonstrated use of NIR-to-visible upconversion fluorescent nanoparticles (UCNPs), emitting visible fluorescence when excited by a NIR light at 980nm, as a nanotransducer for PDT to convert deep tissue-penetrating NIR light to visible light suitable for activating photosensitizers. The unique optical properties of UCNPs enable the upconversion wavelength to be tuned and matched to the activation absorption wavelength of the photosensitizer. At depths beyond 1cm, however, tissue remains inaccessible to light even within the NIR window, and this critical depth limitation renders existing phototherapy ineffective against most deep-seated cancers. We have demonstrated some new treatment modalities for deep-seated cancers based on UCNP hydrogel implants and miniaturized, wirelessly powered optoelectronic devices for light delivery to deep tissues.

Keywords: upconversion, fluorescent, nanoparticle, bioimaging, photodynamic therapy

Procedia PDF Downloads 131

Authors: Tonislav Ivanov, Oleksii Nedashkivskyi, Denis Babeshko, Vadim Pinskiy, Matthew Putman

Abstract:

Training a machine learning model for object detection that generalizes well is known to benefit from a training dataset with diverse examples. However, training datasets usually contain many repeats of common examples of a class and lack rarely seen examples. This is due to the process commonly used during human annotation where a person would proceed sequentially through a list of images labeling a sufficiently high total number of examples. Instead, the method presented involves an active process where, after the initial labeling of several images is completed, the next subset of images for labeling is selected by an algorithm. This process of algorithmic image selection and manual labeling continues in an iterative fashion. The algorithm used for the image selection is a deep learning algorithm, based on the U-shaped architecture, which quantifies the presence of unseen data in each image in order to find images that contain the most novel examples. Moreover, the location of the unseen data in each image is highlighted, aiding the labeler in spotting these examples. Experiments performed using semiconductor wafer data show that labeling a subset of the data, curated by this algorithm, resulted in a model with a better performance than a model produced from sequentially labeling the same amount of data. Also, similar performance is achieved compared to a model trained on exhaustive labeling of the whole dataset. Overall, the proposed approach results in a dataset that has a diverse set of examples per class as well as more balanced classes, which proves beneficial when training a deep learning model.

Keywords: computer vision, deep learning, object detection, semiconductor

Procedia PDF Downloads 112

Authors: Luigi Delle Rose, Oliver Fischer, Ahmed Hammad

Abstract:

In this article, we study the prospects of the proposed Large Hadron electron Collider (LHeC) in the search for heavy neutral scalar particles. We consider a minimal model with one additional complex scalar singlet that interacts with the Standard Model (SM) via mixing with the Higgs doublet, giving rise to an SM-like Higgs boson and a heavy scalar particle. Both scalar particles are produced via vector boson fusion and can be tested via their decays into pairs of SM particles, analogously to the SM Higgs boson. Using multivariate techniques, we show that the LHeC is sensitive to heavy scalars with masses between 200 and 800 GeV down to scalar mixing of order 0.01.

Keywords: beyond the standard model, large hadron electron collider, multivariate analysis, scalar singlet

Procedia PDF Downloads 109

Authors: Bambang Antoro, Lasito Soebari, Geoffrey de Jong, Fernandy Meiriyanto, Michael Siahaan, Eko Wibowo, Pormando Silalahi, Ruswanto, Adi Budirumantyo

Abstract:

The Ertsberg East Skarn System (EESS) is located in the Ertsberg Mining District, Papua, Indonesia. EESS is a sub-vertical zone of copper-gold mineralization hosted in both diorite (vein-style mineralization) and skarn (disseminated and vein style mineralization). Deep Mill Level Zone (DMLZ) is a mining zone in the lower part of East Ertsberg Skarn System (EESS) that product copper and gold. The Deep Mill Level Zone deposit is located below the Deep Ore Zone deposit between the 3125m to 2590m elevation, measures roughly 1,200m in length and is between 350 and 500m in width. DMLZ planned start mined on Q2-2015, being mined at an ore extraction rate about 60,000 tpd by the block cave mine method (the block cave contain 516 Mt). Mineralization and associated hydrothermal alteration in the DMLZ is hosted and enclosed by a large stock (The Main Ertsberg Intrusion) that is barren on all sides and above the DMLZ. Late porphyry dikes that cut through the Main Ertsberg Intrusion are spatially associated with the center of the DMLZ hydrothermal system. DMLZ orebody hosted in diorite and skarn, both dominantly by vein style mineralization. Percentage Material Mined at DMLZ compare with current Reserves are diorite 46% (with 0.46% Cu; 0.56 ppm Au; and 0.83% EqCu); Skarn is 39% (with 1.4% Cu; 0.95 ppm Au; and 2.05% EqCu); Hornfels is 8% (with 0.84% Cu; 0.82 ppm Au; and 1.39% EqCu); and Marble 7 % possible mined waste. Correlation between Ertsberg intrusion, major structure, and vein style mineralization is important to determine characteristic orebody in DMLZ Mine. Generally Deep Mill Level Zone has 2 type of vein filling mineralization from both hosted (diorite and skarn), in diorite hosted the vein system filled by chalcopyrite-bornite-quartz and pyrite, in skarn hosted the vein filled by chalcopyrite-bornite-pyrite and magnetite without quartz. Based on orientation the stockwork vein at diorite hosted and shallow vein in skarn hosted was generally NW-SE trending and NE-SW trending with shallow-moderate dipping. Deep Mill Level Zone control by two main major faults, geologist founded and verified local structure between major structure with NW-SE trending and NE-SW trending with characteristics slickenside, shearing, gauge, water-gas channel, and some has been re-healed.

Keywords: copper-gold, DMLZ, skarn, structure

Procedia PDF Downloads 480

Authors: Amir Mabhout, Toktam Ghafarian, Amirhossein Farzin, Zahra Makki, Sajjad Alizadeh, Amirhossein Ghavi

Abstract:

The high number of Netflix subscribers makes it attractive for data scientists to extract valuable knowledge from the viewers' behavioural analyses. This paper presents a set of statistical insights into viewers' viewing history. After that, a deep learning model is used to predict the future watching behaviour of the users based on previous watching history within the Netflixlatte data pool. Netflixlatte in an aggregated and anonymized data pool of 320 Netflix viewers with a length 250 000 data points recorded between 2008-2022. We observe insightful correlations between the distribution of viewing time and the COVID-19 pandemic outbreak. The presented deep learning model predicts future movie and TV series viewing habits with an average loss of 0.175.

Keywords: data analysis, deep learning, LSTM neural network, netflix

Procedia PDF Downloads 199

Authors: Felix U. Asoiro, Daniel I. Eleazar, Peter O. Offor

Abstract:

As a result of increasing energy demands, research in bioethanol has increased in recent years all through the world, in abide to partially or totally replace renewable energy supplies. The first and third generation feedstocks used for biofuel production have fundamental drawbacks. Waste rice straw and cake from second generation feedstock like Jatropha curcas l. kernel (JC) is seen as non-food feedstock and promising candidates for the industrial production of bioethanol. In this study, JC and rice husk (RH) wastes were characterized for proximate composition. Bioethanol was produced from the residual polysaccharides present in rice husk (RH) and Jatropha seed cake by sequential hydrolytic and fermentative processes at varying mixing proportions (50 g JC/50 g RH, 100 g JC/10 g RH, 100 g JC/20 g RH, 100 g JC/50 g RH, 100 g JC/100 g RH, 100 g JC/200 g RH and 200 g JC/100 g RH) and particle sizes (0.25, 0.5 and 1.00 mm). Mixing proportions and particle size significantly affected both bioethanol yield and some bioethanol properties. Bioethanol yield (%) increased with an increase in particle size. The highest bioethanol (8.67%) was produced at a mixing proportion of 100 g JC/50g RH at 0.25 mm particle size. The bioethanol had the lowest values of specific gravity and density of 1.25 and 0.92 g cm-3 and the highest values of 1.57 and 0.97 g cm-3 respectively. The highest values of viscosity (4.64 cSt) were obtained with 200 g JC/100 g RH, at 1.00 mm particle size. The maximum flash point and cloud point values were 139.9 oC and 23.7oC (100 g JC/200 g RH) at 1 mm and 0.5 mm particle sizes respectively. The maximum pour point value recorded was 3.85oC (100 g JC/50 g RH) at 1 mm particle size. The paper concludes that bioethanol can be recovered from JC and RH wastes. JC and RH blending proportions as well as particle sizes are important factors in bioethanol production.

Keywords: bioethanol, hydrolysis, Jatropha curcas l. kernel, rice husk, fermentation, proximate composition

Procedia PDF Downloads 68

Authors: Ng Mei Han, Nu'man Abdul Hadi

Abstract:

Extraction of ferulic acid from palm pressed fiber using deep eutectic solvent (DES) of choline chloride-acetic acid (ChCl-AA) and choline chloride-citric acid (ChCl-CA) are reported. Influence of water content in DES on the extraction efficiency was investigated. ChCl-AA and ChCl-CA experienced a drop in viscosity from 9.678 to 1.429 and 22.658 ± 1.655 mm2/s, respectively as the water content in the DES increased from 0 to 50 wt% which contributed to higher extraction efficiency for the ferulic acid. Between 41,155 ± 940 mg/kg ferulic acid was obtained after 6 h reflux when ChCl-AA with 30 wt% water was used for the extraction compared to 30,940 ± 621 mg/kg when neat ChCl-AA was used. Although viscosity of the DES could be improved with the addition of water, there is a threshold where the DES could tolerate the presence of water without changing its solvent behavior. The optimum condition for extraction of ferulic acid from palm pressed fiber was heating for 6 h with DES containing 30 wt% water.

Keywords: deep eutectic solvent, extraction, ferulic acid, palm fibre

Procedia PDF Downloads 58

Authors: Syeda Azeem Unnisa

Abstract:

The present study was undertaken at Jawaharnagar Solid Waste Municipal Dumpsite, Greater Hyderabad Municipal Corporation, Telangana State, India in 2017 which generates 90,000 litres of leachate per day. The main objective of the leachate treatment was to remove organic compounds like color, suspended solids, ammonia and COD by coagulation-flocculation using polyaluminum chloride (PAC) as coagulant which has higher coagulant efficiency and relative low cost compared to the conventional coagulants. Jar test apparatus was used to conduct experiments for pH 7, rapid mixing speed 150 rpm for 3 minute, slow mixing speed 30 rpm for 20 minute and the settling time of 30 minute for different dosage of PAC (0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5 and 5.0 g/L). The highest percentage of removal of suspended solids, color, COD and ammonical nitrogen are 97%, 96%, 60% and 37% with PAC optimum dose of 2.0 g/l. The results indicate that the PAC was effective in leachate treatment which is very much suitable for high toxicity of waste and economically feasible for Indian conditions. The treated water can be utilized for other purpose apart from drinking.

Keywords: coagulant, leachate, polyaluminium chloride, treatment

Procedia PDF Downloads 168

Authors: Zabeehullah, Fahim Arif, Yawar Abbas

Abstract:

Software Defined Network (SDN) is a next genera-tion networking model which simplifies the traditional network complexities and improve the utilization of constrained resources. Currently, most of the SDN based Internet of Things(IoT) environments use traditional network routing strategies which work on the basis of max or min metric value. However, IoT network heterogeneity, dynamic traffic flow and complexity demands intelligent and self-adaptive routing algorithms because traditional routing algorithms lack the self-adaptions, intelligence and efficient utilization of resources. To some extent, SDN, due its flexibility, and centralized control has managed the IoT complexity and heterogeneity but still Software Defined IoT (SDIoT) lacks intelligence. To address this challenge, we proposed a model called Deep Routing Strategy (DRS) which uses Deep Learning algorithm to perform routing in SDIoT intelligently and efficiently. Our model uses real-time traffic for training and learning. Results demonstrate that proposed model has achieved high accuracy and low packet loss rate during path selection. Proposed model has also outperformed benchmark routing algorithm (OSPF). Moreover, proposed model provided encouraging results during high dynamic traffic flow.

Keywords: SDN, IoT, DL, ML, DRS

Procedia PDF Downloads 87

Authors: Ilkham Gafurovich Atabaev, Khimmatali Nomozovich Juraev

Abstract:

The activation energy of conductivity in p-i-n SiC structures fabricated by doping with Aluminum using the new low-temperature diffusion method is investigated. In this method, diffusion is stimulated by the flux of carbon and silicon vacancies created by surface oxidation. The activation energy of conductivity in the p - layer is 0.25 eV and it is close to the ionization energy of Aluminum in 4H-SiC from 0.21 to 0.27 eV for the hexagonal and cubic positions of aluminum in the silicon sublattice for weakly doped crystals. The conductivity of the i-layer (measured in the reverse biased diode) shows 2 activation energies: 0.02 eV and 0.62 eV. Apparently, the 0.62 eV level is a deep trap level and it is a complex of Aluminum with a vacancy. According to the published data, an analogous level system (with activation energies of 0.05, 0.07, 0.09 and 0.67 eV) was observed in the ion Aluminum doped 4H-SiC samples.

Keywords: activation energy, aluminum, low temperature diffusion, SiC

Procedia PDF Downloads 250

Authors: Mst Shapna Akter, Hossain Shahriar

Abstract:

One of the most important challenges in the field of software code audit is the presence of vulnerabilities in software source code. Every year, more and more software flaws are found, either internally in proprietary code or revealed publicly. These flaws are highly likely exploited and lead to system compromise, data leakage, or denial of service. C and C++ open-source code are now available in order to create a largescale, machine-learning system for function-level vulnerability identification. We assembled a sizable dataset of millions of opensource functions that point to potential exploits. We developed an efficient and scalable vulnerability detection method based on deep neural network models that learn features extracted from the source codes. The source code is first converted into a minimal intermediate representation to remove the pointless components and shorten the dependency. Moreover, we keep the semantic and syntactic information using state-of-the-art word embedding algorithms such as glove and fastText. The embedded vectors are subsequently fed into deep learning networks such as LSTM, BilSTM, LSTM-Autoencoder, word2vec, BERT, and GPT-2 to classify the possible vulnerabilities. Furthermore, we proposed a neural network model which can overcome issues associated with traditional neural networks. Evaluation metrics such as f1 score, precision, recall, accuracy, and total execution time have been used to measure the performance. We made a comparative analysis between results derived from features containing a minimal text representation and semantic and syntactic information. We found that all of the deep learning models provide comparatively higher accuracy when we use semantic and syntactic information as the features but require higher execution time as the word embedding the algorithm puts on a bit of complexity to the overall system.

Keywords: cyber security, vulnerability detection, neural networks, feature extraction

Procedia PDF Downloads 56

Authors: Chiau Yuan Lim, Hayyiratul Fatimah Mohd Zaid, Fai Kait Chong

Abstract:

Deep Eutectic Solvent (DES) is used in various applications due to its simplicity in synthesis procedure, biodegradable, inexpensive and easily available chemical ingredients. Graphene Oxide is a popular catalyst that being used in various processes due to its stacking carbon sheets in layer which theoretically rapid up the catalytic processes. In this study, choline chloride based DESs were synthesized and ChCl-PEG(1:4) was found to be the most effective DES in performing desulfurization, which it is able to remove up to 47.4% of the sulfur content in the model oil in just 10 minutes, and up to 95% of sulfur content after repeat the process for six times. ChCl-PEG(1:4) able to perform up to 32.7% desulfurization on real diesel after 6 multiple stages. Thus, future research works should focus on removing the impurities on real diesel before utilising DESs in petroleum field.

Keywords: choline chloride, deep eutectic solvent, fuel desulfurization, graphene oxide

Procedia PDF Downloads 126

Authors: Hitendra Raj Joshi, Bipin Lamichhane

Abstract:

Groundwater is the main source of water for the industries of Balaju Industrial District (BID) and the denizens of Balaju-Boratar area. The quantity of groundwater is in a fatal condition in the area than earlier days. Water levels in shallow wells have highly lowered and deep wells are not providing an adequate amount of water as before because of higher extraction rate than the recharge rate. The main recharge zone of the shallow aquifer lies at the foot of Nagarjuna mountain, where recent colluvial debris are accumulated. Urbanization in the area is the main reason for decreasing water table. Recharge source for the deep aquifer in the region is aquiclude leakage. Sand layer above the Kalimati clay is the shallow aquifer zone, which is limited only in Balaju and eastern part of the Boratar, while the layer below the Kalimati clay spreading around Gongabu, Machhapohari, and Balaju area is considered as a potential area of deep aquifer. Over extraction of groundwater without considering water balance in the aquifers may dry out the source and can initiate the land subsidence problem. Hence, all the responsible of the industries in BID area and the denizens of Balaju-Boratar area should be encouraged to practice artificial groundwater recharge.

Keywords: aquiclude leakage, Kalimati clay, groundwater recharge

Procedia PDF Downloads 471

Authors: Yukiko Sasaki Alam, Shahid Alam

Abstract:

Syntactic parsing is vital for semantic treatment by many applications related to natural language processing (NLP), because form and content coincide in many cases. However, it has not yet reached the levels of reliable performance. By manually examining and analyzing individual machine translation output errors that involve syntax as well as semantics, this study attempts to discover what is required for improving syntactic and semantic parsing.

Keywords: syntactic parsing, error analysis, machine translation, deep parsing

Procedia PDF Downloads 525

Authors: Ziyda Abunada

Abstract:

In-situ remediation of contaminated land with deep mixing can deliver a multi-technique remedial strategy. A field trail includes permeable reactive barrier (PRB) took place at a severely contaminated site in Yorkshire to the north of the UK through the SMiRT (Soil Mix Remediation Technology) project in May 2011. SMiRT involved the execution of the largest research field trials in the UK to provide field validation. Innovative modified bentonite materials in combination with zeolite and organoclay were used to construct six different walls of a hexagonal PRB. Field monitoring, testing and site cores were collected from the PRB twice: once 2 months after the construction and again in March 2014 (almost 34 months later).This paper presents an overview of the results of the PRB materials’ relative performance with some initial 3-year time-related assessment. Results from the monitoring program and the site cores are presented. Some good correlations are seen together with some clear difference among the materials’ efficiency. These preliminary observations represent a potential for further investigations and highlighted the main lessons learned in a filed scale.

Keywords: in-situ remediation, groundwater, permeable reactive barrier, site cores

Procedia PDF Downloads 182

Authors: James A. Adeyanju, John O. Olajide, Akinbode A. Adedeji

Abstract:

A predictive mathematical model based on the fundamental principles of mass transfer was developed to simulate the moisture content and oil content during Deep-Fat Frying (DFF) process of dodo. The resulting governing equation, that is, partial differential equation that describes rate of moisture loss and oil uptake was solved numerically using explicit Finite Difference Technique (FDT). Computer codes were written in MATLAB environment for the implementation of FDT at different frying conditions and moisture loss as well as oil uptake simulation during DFF of dodo. Plantain samples were sliced into 5 mm thickness and fried at different frying oil temperatures (150, 160 and 170 ⁰C) for periods varying from 2 to 4 min. The comparison between the predicted results and experimental data for the validation of the model showed reasonable agreement. The correlation coefficients between the predicted and experimental values of moisture and oil transfer models ranging from 0.912 to 0.947 and 0.895 to 0.957, respectively. The predicted results could be further used for the design, control and optimization of deep-fat frying process.

Keywords: frying, moisture loss, modelling, oil uptake

Procedia PDF Downloads 413

Authors: Yujiang Wu

Abstract:

As electrical machine (e-machine) power density re-querulents become more stringent in vehicle electrification, mounting a temperature sensor for e-machine stator windings becomes increasingly difficult. This can lead to higher manufacturing costs, complicated harnesses, and reduced reliability. In this paper, we propose a deep-learning method for predicting electric machine winding temperature, which can either replace the sensor entirely or serve as a backup to the existing sensor. We compare the performance of our method, the stateful long short-term memory networks (LSTM) with truncated backpropagation through time (TBTT), with that of linear regression, as well as stateless LSTM with/without residual connection. Our results demonstrate the strength of combining stateful LSTM and TBTT in tackling nonlinear time series prediction problems with long sequence lengths. Additionally, in industrial applications, high-temperature region prediction accuracy is more important because winding temperature sensing is typically used for derating machine power when the temperature is high. To evaluate the performance of our algorithm, we developed a temperature-stratified MSE. We propose a simple but effective data preprocessing trick to improve the high-temperature region prediction accuracy. Our experimental results demonstrate the effectiveness of our proposed method in accurately predicting winding temperature, particularly in high-temperature regions, while also reducing manufacturing costs and improving reliability.

Keywords: deep learning, electrical machine, functional safety, long short-term memory networks (LSTM), thermal management, time series prediction

Procedia PDF Downloads 64

Authors: Majid Hejazian, Nam-Trung Nguyen

Abstract:

Over the past few years, microfluidic devices have generated significant attention from industry and academia due to advantages such as small sample volume, low cost and high efficiency. Microfluidic devices have applications in chemical, biological and industry analysis and can facilitate assay of bio-materials and chemical reactions, separation, and sensing. Micromixers are one of the important microfluidic concepts. Micromixers can work as stand-alone devices or be integrated in a more complex microfluidic system such as a lab on a chip (LOC). Micromixers are categorized as passive and active types. Passive micromixers rely only on the arrangement of the phases to be mixed and contain no moving parts and require no energy. Active micromixers require external fields such as pressure, temperature, electric and acoustic fields. Rapid and efficient mixing is important for many applications such as biological, chemical and biochemical analysis. Achieving fast and homogenous mixing of multiple samples in the microfluidic devices has been studied and discussed in the literature recently. Improvement in mixing rely on effective mass transport in microscale, but are currently limited to molecular diffusion due to the predominant laminar flow in this size scale. Using magnetic field to elevate mass transport is an effective solution for mixing enhancement in microfluidics. The use of a non-uniform magnetic field to improve mass transfer performance in a microfluidic device is demonstrated in this work. The phenomenon of mixing ferrofluid and DI-water streams has been reported before, but mass transfer enhancement for other non-magnetic species through magnetic field have not been studied and evaluated extensively. In the present work, permanent magnets were used in a simple microfluidic device to create a non-uniform magnetic field. Two streams are introduced into the microchannel: one contains fluorescent dye mixed with diluted ferrofluid to induce enhanced mass transport of the dye, and the other one is a non-magnetic DI-water stream. Mass transport enhancement of fluorescent dye is evaluated using fluorescent measurement techniques. The concentration field is measured for different flow rates. Due to effect of magnetic field, a body force is exerted on the paramagnetic stream and expands the ferrofluid stream into non-magnetic DI-water flow. The experimental results demonstrate that without a magnetic field, both magnetic nanoparticles of the ferrofluid and the fluorescent dye solely rely on molecular diffusion to spread. The non-uniform magnetic field, created by the permanent magnets around the microchannel, and diluted ferrofluid can improve mass transport of non-magnetic solutes in a microfluidic device. The susceptibility mismatch between the fluids results in a magnetoconvective secondary flow towards the magnets and subsequently the mass transport of the non-magnetic fluorescent dye. A significant enhancement in mass transport of the fluorescent dye was observed. The platform presented here could be used as a microfluidics-based micromixer for chemical and biological applications.

Keywords: ferrofluid, mass transfer, micromixer, microfluidics, magnetic

Procedia PDF Downloads 196

Authors: Leila Keshavarz Afshar, Hedieh Sajedi

Abstract:

Estimation of biological brain age from MR images is a topic that has been much addressed in recent years due to the importance it attaches to early diagnosis of diseases such as Alzheimer's. In this paper, we use a 3D Convolutional Neural Network (CNN) to provide a method for estimating the biological age of the brain. The 3D-CNN model is trained by MRI data that has been normalized. In addition, to reduce computation while saving overall performance, some effectual slices are selected for age estimation. By this method, the biological age of individuals using selected normalized data was estimated with Mean Absolute Error (MAE) of 4.82 years.

Keywords: brain age estimation, biological age, 3D-CNN, deep learning, T1-weighted image, SPM, preprocessing, MRI, canny, gray matter

Procedia PDF Downloads 121

Authors: A. M. Gsiea, J. P. Goss, P. R. Briddon, Ramadan. M. Al-habashi, K. M. Etmimi, Khaled. A. S. Marghani

Abstract:

Using first-principles methods based on density functional theory and pseudopotentials, we have performed a details study of native defects in ZnO. Native point defects are unlikely to be cause of the unintentional n-type conductivity. Oxygen vacancies, which considered most often been invoked as shallow donors, have high formation energies in n-type ZnO, in edition are a deep donors. Zinc interstitials are shallow donors, with high formation energies in n-type ZnO, and thus unlikely to be responsible on their own for unintentional n-type conductivity under equilibrium conditions, as well as Zn antisites which have higher formation energies than zinc interstitials. Zinc vacancies are deep acceptors with low formation energies for n-type and in which case they will not play role in p-type coductivity of ZnO. Oxygen interstitials are stable in the form of electrically inactive split interstitials as well as deep acceptors at the octahedral interstitial site under n-type conditions. Our results may provide a guide to experimental studies of point defects in ZnO.

Keywords: DFT, native, n-type, ZnO

Procedia PDF Downloads 561

Authors: Praveen Chandramenon, Andrew Gascoyne, Fideline Tchuenbou-Magaia

Abstract:

Aquaponics offers a simple conclusive solution to the food and environmental crisis of the world. This approach combines the idea of Aquaculture (growing fish) to Hydroponics (growing vegetables and plants in a soilless method). Smart Aquaponics explores the use of smart technology including artificial intelligence and IoT, to assist farmers with better decision making and online monitoring and control of the system. Identification of different growth stages of Swiss Chard plants and predicting its harvest time is found to be important in Aquaponic yield management. This paper brings out the comparative analysis of a standard Aquaponics with a Vermiponics (Aquaponics with worms), which was grown in the controlled environment, by implementing IoT and deep learning-based growth stage segregation and harvest time prediction of Swiss Chards before and after applying an optimal freshwater replenishment. Data collection, Growth stage classification and Harvest Time prediction has been performed with and without water replenishment. The paper discusses the experimental design, IoT and sensor communication with architecture, data collection process, image segmentation, various regression and classification models and error estimation used in the project. The paper concludes with the results comparison, including best models that performs growth stage segregation and harvest time prediction of the Aquaponic and Vermiponic testbed with and without freshwater replenishment.

Keywords: aquaponics, deep learning, internet of things, vermiponics

Procedia PDF Downloads 45

Authors: Hailyie Tekleselase

Abstract:

Deep learning is increasingly used as a building block of security systems. However, neural networks are hard to interpret and typically solid to the practitioner. This paper presents a detail survey of computing methods in cyber security, and analyzes the prospects of enhancing the cyber security capabilities by suggests that of accelerating the intelligence of the security systems. There are many AI-based applications used in industrial scenarios such as Internet of Things (IoT), smart grids, and edge computing. Machine learning technologies require a training process which introduces the protection problems in the training data and algorithms. We present machine learning techniques currently applied to the detection of intrusion, malware, and spam. Our conclusions are based on an extensive review of the literature as well as on experiments performed on real enterprise systems and network traffic. We conclude that problems can be solved successfully only when methods of artificial intelligence are being used besides human experts or operators.

Keywords: artificial intelligence, machine learning, deep learning, cyber security, big data

Procedia PDF Downloads 102

Authors: Edward Holupka, John Rossman, Tye Morancy, Joseph Aronovitz, Irving Kaplan

Abstract:

A common modality for the treatment of early stage prostate cancer is the implantation of radioactive seeds directly into the prostate. The radioactive seeds are positioned inside the prostate to achieve optimal radiation dose coverage to the prostate. These radioactive seeds are positioned inside the prostate using Transrectal ultrasound imaging. Once all of the planned seeds have been implanted, two dimensional transaxial transrectal ultrasound images separated by 2 mm are obtained through out the prostate, beginning at the base of the prostate up to and including the apex. A common deep neural network, called DetectNet was trained to automatically determine the position of the implanted radioactive seeds within the prostate under ultrasound imaging. The results of the training using 950 training ultrasound images and 90 validation ultrasound images. The commonly used metrics for successful training were used to evaluate the efficacy and accuracy of the trained deep neural network and resulted in an loss_bbox (train) = 0.00, loss_coverage (train) = 1.89e-8, loss_bbox (validation) = 11.84, loss_coverage (validation) = 9.70, mAP (validation) = 66.87%, precision (validation) = 81.07%, and a recall (validation) = 82.29%, where train and validation refers to the training image set and validation refers to the validation training set. On the hardware platform used, the training expended 12.8 seconds per epoch. The network was trained for over 10,000 epochs. In addition, the seed locations as determined by the Deep Neural Network were compared to the seed locations as determined by a commercial software based on a one to three months after implant CT. The Deep Learning approach was within \strikeout off\uuline off\uwave off2.29\uuline default\uwave default mm of the seed locations determined by the commercial software. The Deep Learning approach to the determination of radioactive seed locations is robust, accurate, and fast and well within spatial agreement with the gold standard of CT determined seed coordinates.

Keywords: prostate, deep neural network, seed implant, ultrasound

Procedia PDF Downloads 170

Authors: Vijaya Yuvaram Singh V M, Kameshwar Rao J V

Abstract:

The challenge with development of neural network based methods for medical is the availability of data. Anatomical landmark detection in the medical domain is a process to find points on the x-ray scan report of the patient. Most of the time this task is done manually by trained professionals as it requires precision and domain knowledge. Traditionally object detection based methods are used for landmark detection. Here, we utilize reinforcement learning and query based method to train a single agent capable of detecting multiple landmarks. A deep Q network agent is trained to detect single and multiple landmarks present on hip and shoulder from x-ray scan of a patient. Here a single agent is trained to find multiple landmark making it superior to having individual agents per landmark. For the initial study, five images of different patients are used as the environment and tested the agents performance on two unseen images.

Keywords: reinforcement learning, medical landmark detection, multi target detection, deep neural network

Procedia PDF Downloads 122

Authors: Jiaxin Shi, Kaifeng Hao, Qingfan An, Zeng Peng

Abstract:

Due to a lack of data sources and data processing techniques, it has always been difficult to quantify public space quality, which includes urban construction quality and how it is perceived by people, especially in large urban areas. This study proposes a quantitative research method based on the consideration of emotional health and physical health of the built environment. It highlights the low quality of public areas in Tianjin, China, where there are many negative elements. Deep learning technology is then used to measure how effectively people perceive urban areas. First, this work suggests a deep learning model that might simulate how people can perceive the quality of urban construction. Second, we perform semantic segmentation on street images to identify visual elements influencing scene perception. Finally, this study correlated the scene perception score with the proportion of visual elements to determine the surrounding environmental elements that influence scene perception. Using a small-scale labeled Tianjin street view data set based on transfer learning, this study trains five negative spatial discriminant models in order to explore the negative space distribution and quality improvement of urban streets. Then it uses all Tianjin street-level imagery to make predictions and calculate the proportion of negative space. Visualizing the spatial distribution of negative space along the Tianjin Inner Ring Road reveals that the negative elements are mainly found close to the five key districts. The map of Tianjin was combined with the experimental data to perform the visual analysis. Based on the emotional assessment, the distribution of negative materials, and the direction of street guidelines, we suggest guidance content and design strategy points of the negative phenomena in Tianjin street space in the two dimensions of perception and substance. This work demonstrates the utilization of deep learning techniques to understand how people appreciate high-quality urban construction, and it complements both theory and practice in urban planning. It illustrates the connection between human perception and the actual physical public space environment, allowing researchers to make urban interventions.

Keywords: human perception, public space quality, deep learning, negative elements, street images

Procedia PDF Downloads 82

Authors: Chinlun Lai, Lunjyh Jiang

Abstract:

Due to the heavy burden and pressure of caring for infants, an integrated automatic baby watching system based on IoT smart sensing and deep learning machine vision techniques is proposed in this paper. By monitoring infant body conditions such as heartbeat, breathing, body temperature, sleeping posture, as well as the surrounding conditions such as dangerous/sharp objects, light, noise, humidity and temperature, the proposed system can analyze and predict the obvious/potential dangerous conditions according to observed data and then adopt suitable actions in real time to protect the infant from harm. Thus, reducing the burden of the caregiver and improving safety efficiency of the caring work. The experimental results show that the proposed system works successfully for the infant care work and thus can be implemented in various life fields practically.

Keywords: baby care system, Internet of Things, deep learning, machine vision

Procedia PDF Downloads 207

Authors: Xiaocong Liu, Huazhen Wang, Ting He, Xiaozheng Li, Weihan Zhang, Jian Chen

Abstract:

The utilization of electronic medical record (EMR) data to establish the disease diagnosis model has become an important research content of biomedical informatics. Deep learning can automatically extract features from the massive data, which brings about breakthroughs in the study of EMR data. The challenge is that deep learning lacks semantic knowledge, which leads to impracticability in medical science. This research proposes a method of incorporating lexical-semantic knowledge from abundant entities into a convolutional neural network (CNN) framework for pediatric disease diagnosis. Firstly, medical terms are vectorized into Lexical Semantic Vectors (LSV), which are concatenated with the embedded word vectors of word2vec to enrich the feature representation. Secondly, the semantic distribution of medical terms serves as Semantic Decision Guide (SDG) for the optimization of deep learning models. The study evaluate the performance of LSV-SDG-CNN model on four kinds of Chinese EMR datasets. Additionally, CNN, LSV-CNN, and SDG-CNN are designed as baseline models for comparison. The experimental results show that LSV-SDG-CNN model outperforms baseline models on four kinds of Chinese EMR datasets. The best configuration of the model yielded an F1 score of 86.20%. The results clearly demonstrate that CNN has been effectively guided and optimized by lexical-semantic knowledge, and LSV-SDG-CNN model improves the disease classification accuracy with a clear margin.

Keywords: convolutional neural network, electronic medical record, feature representation, lexical semantics, semantic decision

Procedia PDF Downloads 111

Authors: M. Agostini, A. Sonato, G. Greco, M. Travagliati, G. Ruffato, E. Gazzola, D. Liuni, F. Romanato, M. Cecchini

Abstract:

Since their first demonstration in the early 1980s, surface plasmon resonance (SPR) sensors have been widely recognized as useful tools for detecting chemical and biological species, and the interest of the scientific community toward this technology has known a rapid growth in the past two decades owing to their high sensitivity, label-free operation and possibility of real-time detection. Recent works have suggested that a turning point in SPR sensor research would be the combination of SPR strategies with other technologies in order to reduce human handling of samples, improve integration and plasmonic sensitivity. In this light, microfluidics has been attracting growing interest. By properly designing microfluidic biochips it is possible to miniaturize the analyte-sensitive areas with an overall reduction of the chip dimension, reduce the liquid reagents and sample volume, improve automation, and increase the number of experiments in a single biochip by multiplexing approaches. However, as the fluidic channel dimensions approach the micron scale, laminar flows become dominant owing to the low Reynolds numbers that typically characterize microfluidics. In these environments mixing times are usually dominated by diffusion, which can be prohibitively long and lead to long-lasting biochemistry experiments. An elegant method to overcome these issues is to actively perturb the liquid laminar flow by exploiting surface acoustic waves (SAWs). With this work, we demonstrate a new approach for SPR biosensing based on the combination of microfluidics, SAW-induced mixing and the real-time phase-interrogation grating-coupling SPR technology. On a single lithium niobate (LN) substrate the nanostructured SPR sensing areas, interdigital transducer (IDT) for SAW generation and polydimethylsiloxane (PDMS) microfluidic chambers were fabricated. SAWs, impinging on the microfluidic chamber, generate acoustic streaming inside the fluid, leading to chaotic advection and thus improved fluid mixing, whilst analytes binding detection is made via SPR method based on SPP excitation via gold metallic grating upon azimuthal orientation and phase interrogation. Our device has been fully characterized in order to separate for the very first time the unwanted SAW heating effect with respect to the fluid stirring inside the microchamber that affect the molecules binding dynamics. Avidin/biotin assay and thiol-polyethylene glycol (bPEG-SH) were exploited as model biological interaction and non-fouling layer respectively. Biosensing kinetics time reduction with SAW-enhanced mixing resulted in a ≈ 82% improvement for bPEG-SH adsorption onto gold and ≈ 24% for avidin/biotin binding—≈ 50% and 18% respectively compared to the heating only condition. These results demonstrate that our biochip can significantly reduce the duration of bioreactions that usually require long times (e.g., PEG-based sensing layer, low concentration analyte detection). The sensing architecture here proposed represents a new promising technology satisfying the major biosensing requirements: scalability and high throughput capabilities. The detection system size and biochip dimension could be further reduced and integrated; in addition, the possibility of reducing biological experiment duration via SAW-driven active mixing and developing multiplexing platforms for parallel real-time sensing could be easily combined. In general, the technology reported in this study can be straightforwardly adapted to a great number of biological system and sensing geometry.

Keywords: biosensor, microfluidics, surface acoustic wave, surface plasmon resonance

Procedia PDF Downloads 248