Conflict-Receptive and Prognosis Scheduling in Deep Learning Systems

Irish Interdisciplinary Journal of Science & Research (IIJSR)
Vol.6, Iss.2, Pages 43-55, April-June 2022
ISSN: 2582-3981 www.iijsr.com
43
Conflict-Receptive and Prognosis Scheduling in Deep Learning Systems
K.Sathyalakshmi1
, Dr.T.Ananth Kumar2
, N.Padma Priya3
& S.Jayalakshmi4
1-4
IFET College of Engineering, Villupuram-605108, Tamilnadu, India.
DOI: http://doi.org/10.46759/IIJSR.2022.6207
Copyright © 2022 K.Sathyalakshmi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Article Received: 19 February 2022 Article Accepted: 26 April 2022 Article Published: 31 May 2022
1. Introduction
As in many other businesses, there is a tremendous drive in theelectronics manufacturing industry to
become more environmentally friendly. The appetite for the most popular electronic equipment has been
increasing for several years and will continue to surge for even more. Since carbon is one of the most
ubiquitous and needed commodities in consumer electronics, tackling this challenge's carbon footprint is now
a main consideration, if not a problem, for manufactures. Marketing Analysis estimates that the electronic
component business will develop by 5.6 percent over the next five years [1]. In many industrial companies,
consumables are manufactured in huge quantities, and alternative products may be created consecutively. The
produced products are then delivered to a variety of dealers, from them customers may acquire the quality
products. Manufacturers may assess product sales and demand from dealers who are the closest point of
contact with clients, and from that information they can discern which product is in short supply and allowing
them to expand manufacturing of that goods in a continuously running operation [2].
Electronic products industry is advancing at a compound annual growth rate (CAGR) of 22% and is
forecasted to reach $400 billion USD by 2025 [3]. The Indian government has also taken many initiatives to
strengthen production in this domain, positioning India at the top of the list of speculative investment regions.
The flow of consecutive stages is crucial in mass production, with the accomplishment of one step prompting
the commencement of the next [4]. To achieve optimum efficiency, process manufacturers generally depend
on monitoring and tracking tools and applications. They cannot schedule for constantly changing data in the
current framework since our concept is built on electronic manufacturing businesses, and there is accelerating
need for electronics as we live in the internet world.
ABSTRACT
In Manufacturing Industry, the main process involved are product design, prototype, product testing and then finally mass production is
instantiated. So, most of the manufacturing industries and their production processes are static where the process is not flexible enough to
handle sudden changes in production mechanism. In our case, consider the electronic manufacturing industry they undergo through the same
phase where they are not capable of sudden changes in the production process, sudden changes in the sense where manufacturing company
must analyze on demand products then try to satisfy the customer demand. In previous model static manufacturing process is handled by using
FIFO algorithm which helps to schedule the process, in which the process arrives first will be executed first and it is a non-pre-emptive type of
scheduling algorithm. The existing is not suitable to solve a dynamic manufacturing process where the scheduling should also be dynamic
according to the priority. So, we proposed a Priority scheduling algorithm, and it is pre-emptive type, eventually based on scheduling, our
proposed algorithm helps to schedule any process depending upon the priority. The proposed algorithm helps to schedule and also it
dynamically change the process depending on the products priority, where one process may have to stop so the process of priority products
can be initiated. By achieving the dynamic manufacturing process any business can be developed as they can satisfy the on-demand nature and
their new product production process concurrently.
Keywords: Manufacturing Industry, FIFO Algorithm, Non-pre-emptive Type, Priority Scheduling Algorithm, Priority, Pre-emptive Type,
Dynamic Manufacturing.

44
The electronics manufacturing industry is competitive. Electronics manufacturing is evolved from electronics
production service [5]. As a result, electronics production is a type of subcontracting division. Additionally, it
offers a diverse range of central assembly proportions. As a result, the demand for contracted parallel tasks
and circuit boards is booming. Based on the organization and product, different levels of automation are used
in electronic production. Companies that make vast quantities of items typically employ fully automated
technology [6]. Numerous firms build and evaluate electronic goods in order to constructing them. Electronic
manufacturing businesses may specialize in one or even more sectors of production process and production
size. The electronic industry creates equipment and electronic goods as well as electrical components for a
broad spectrum of products [7],[8]. Mobile devices and notebook computers are prominent goods in the
semiconductor industry. The semiconductor industry changed workplaces, businesses, and dwellings,
positioning itself as a massive macroeconomic component on track with the chemical, steel, and automobile
economies in strength [9].
Manufacturing is the large-scale preparation and assembly of components and end wares. It can use
equipment to transform raw resources into end products employing a variety of techniques, comprising
people and machines activity, as well as microbiological operations. The employment of novel materials,
organic electronics, and simplification are among the key developments in manufacturing parts [10].
Production is similar to manufacturing but has a bigger appeal. It relates to the method and procedures used to
turn crude ingredients or semi-finished materials into completed goods or services, whether the machines is
incorporated or not. Manufacturers must adapt their production methods to the economy's demands and
wishes, available resources, order frequency and volume, seasonal fluctuations in demand, overhead
expenses, and a variety of other factors [11]. Electronics assembling is a broad term for the process of
assembling, welding, or merging electronic components and circuits to fulfill one or more specific actions. It
is a necessary component in the formation of daily electronic equipment such as laptops, gadgets,
automobiles, remote controls, and phones.
So, our suggested algorithm, Priority Scheduling, aids our process in modifying its character in accordance to
ongoing demand and concentrating on contemporary research and products that come to market. It is crucial
to measure all of the constituents of the substances used by the producer to buildan electronic gadget during
the design process. Before manufacture, the equipment is tested to ensure it is working correctly under a
diverse setting [12]. Expert designers include a variety of related tasks that take place after determining a
user's requirements or product requirements but before production or installation takes place. This is critical
because if there is even a tiny error in the design, the entire manufacturing will be a waste of effort. As a
result, you should focus on one particular item; examine it, and then go on to commercial manufacturing.
Production is amongst the most global ambitions in production and is a fundamental component. There would
be no finished commodities to sell to buyers if this activity did not take place. As a result, manufacturing
contributes so much to improvements in living standards than any other industry [13]. Once the process is
finished with the design and testing, it’s time to disseminate them. However, before you begin delivery, do
a few last checks to assure that your clients do not receive faulty items. This will ensures that the reputation

45
of your goods or organization is not tarnished. A thriving industrial base lead to high R&D, invention,
profitability, and exporting [14]. Developed in order to greater financial action than other sectors. It also
promotes good inter-industry and cross - functional and cross ties, business productivity, technical
development, and invention. Thus, our algorithm aids in the scheduling of continuously shifting activities,
which not only enhances performance but also boosts economic activity through client fulfillment and the
creation of a healthy stock chain.
2. Literature Survey
Many existing deep learning systems analyze workflows to optimize resource consumption; critical factors
comprise learning state communication styles, engine scheduling designs, and inferred processing time.
Concerning GPU consumption benchmarking, it relies on period and leverages continuous profiling on
segregated workstations to establish effective co- location and relocation options. This consistency is also
performed to examine employee productivity and proactively relocate jobs to better-fit Processors, enhancing
cluster productivity. The prototype model illustrates that processes can be delayed and relocated in just a
moment, including for cross-server conversion for popular neural network tools and features. It dramatically
improves cluster productivity indicating a novel framework for managing huge GPU ensembles for transfer
learning [15].
This paradigm is intended to complement standard methods to understand distinct software systems. To serve
more intricate deep learning algorithms, new specific training technology, application layers, and refinement
techniques are being developed [16]. As a result, there is an acute need to establish a proper scientific strategy
for detailed analysis software of deep learning- optimized hardware and software systems. This article does
not address deep learning induction, cloud usage, cross configurations, efficiency, scalability, or other
computational intelligence frameworks [17]. It forecasts time for training based on numerical characteristics,
machine parameters, and per-layer query execution stereotyping. As a result of heterogeneity in embedded
system, demand, and long wait times, DL-specific group queues are necessary [18].
It makes use of excess Graphics hardware to co-executer numerousprocesses on a single Device. In modern
operating systems and linked CPU cores, economic seclusion is fundamental. The goal of DL jobs is to
enhance typical and job depletion period. It manages this by monitoring network delay, consolidating global
DL jobs, and building a multi-level feed buffer that modulates job priorities. This permits for fine-grained
related to process while eliminating activity disruption. It doubles global GPU memory use by 42 percent in
terms and computation usage by 34% in our multi-tenant ensemble without diminishing equity,
demonstrating a unique concept to fully exploit GPUs at scale [19]. Because these GPU capacity coordinators
operate between the GPU controller and the runtime environment, they are unable to organizeand maximize
with a wider perspective. It proposes a method for predicting slowness via co-located DL operations
leveraging emulated GPU statistics in confinement. DL workload and facility aspects are also used to
evaluate a reasonable workout routine and GPU consumption is used to suggest applications that may be
feasible for co- location [20]. The potential explanations design provides use of our models to schedule

46
Machines in attempt to lessen disruption. Tests show that our conclusions boost validity of the model (by 15%
to 40%) and that the planner decreases software run-time latency by 24.2 percent in realistic cases.
Conversely, some group tasks that mitigate performance interference of mixed tasks in cloud platforms are
not equipped to handle DL cluster scheduling successfully.
The proposed solution [21] is to feature activities in GPU hardware, which consumes designs for just a few
seconds; nevertheless, this is unsuitable for DL operations, as DL workloads generally necessitate data
pre-processing whenever the micro chunk is retrieved from the DFS, which might also taketens of minutes.
It was intended to serve as an intrusion controller at the middle levels, decreasing disruption amongst
co-located Machines. It incorporates a throughput prediction model, which adapts the lot of significant factor
servers or users at execution. It presumes that saturation is foreseeable, which is not always the scenario
[22]-[24]. It extends the foundation DL infrastructure and provide perfectly alright kernel rescheduling for
co-location to eliminate noise, because it still involves performance assessment. All of the aforementioned
DL memory utilization supervisors are beneficial to our task since they depend on specific constraints and
programming ambitions [25]. Horus enhances on earlier study by emphasizing on DL network and GPU
consumption.
3. Existing System
First in, first out (FIFO) is a scheduling and non-preemptive strategy that works queued demands and operates
in the sequence in which they arise. It is the most basic and straight forward CPU scheduling technique. This
sort of procedure demands the task first and then assigns the session. A FIFO queue is used to accomplish this.
In the reality, FIFO ordering would be comparable to purchasing a ticket at a ticket counter. As we all know,
while buying a ticket, each passenger is treated in a lineup manner. Likewise, the same method applies for
FIFO processing.
Thus, the process that reaches first in a sequence is executed initially,followed by the processes that continue.
What comes in first must go out first, according to the FIFO methodology. For example, if a batch of 1,000
things is manufactured during the first week of a month and another batch of 1,000 items is made in the
second week, the batch produced initially is auctioned early. Companies that run on the first in, first out
principle rate stocks on the presumption that the very first items ordered will be the first goods manufactured.
This may not be possible under certain instances. Normally, the organization can enact the workflow using
FIFO order, and the manufacturing process is set to begin and implementing the methodology without any
disruptions; however, if any merchandise insisted the operation, it would not halt, and the demand product
would have to wait untilthe preceding process finished its manufacturing process. Because of the fluctuations
in the economics and the likelihood that the production of things will increase over time, this kind of operation
increases the demandsubstantially, and the business confronts many challenges such as timing differences and
breakdowns during production.
From below diagram, we can see that the product which is ordered first will be manufactured. So, the
demanded product must wait for some periodof time until the previous products ordered by dealers have to be

47
manufactured. This will make time delay for the demand products which makes the dealers to deal with the
unhappy customers.
Fig. 3.1. FIFO Manufacturing Process
4. Proposed System
"Production is the systematic process of changing resources into completed products and services; the
objective of manufacturing is to increase production for such transformed resources." Scheduling is one of the
most relevant information of the system administrator since it specifies which tasks are in the waiting list.
Priority scheduling is a pre-emptive method that is one of the most commonly used in batch systems. The
process with the highest priority will be carried out first, and so on. Processes with the same priority are
performed in the order in which they were received by first come first serve (FCFS) basis. It is commonly
used in operating systems for repetitive tasks. Priority scheduling algorithms are classified into two groups.
One is Preemptive priority scheduling while the other is Non-Preemptive Priority scheduling. The priority
number assigned to each of the process may or may not vary.
In priority scheduling, each process is allocated a number that represents its priority level. Lower is the
number, higher is the priority. Unlike other methods of scheduling, the scheduler decides which tasks to work
on based on their importance. The determination of the priority of the tasks is the key to employing
priority-based preemption. The priority of a task is a numeric number assigned to it that is used to determine
which of the tasks that are now available to run will be executed.
The highest-priority task "wins" and the other task remains in the ready queue until the higher priority task
completes or blocks. Priority can be decided based on memory requirements, time requirements or any other
resource requirement like our consumers on demand products. When the company manufactures various
kinds of products in scheduled manner, priority-based products may have to wait for the previous process to
complete. Above problem can be mitigated while using Priority Scheduling algorithm where it can be flexible
when it comes to processing technique where priority-based products can be instantiated first then second
priority products process is initialized. In this method, the scheduler chooses the tasks to work as per the
priority, which is different from other types of scheduling.

48
The emphasis will be given to the goods based on dealership requests. So, in our example, the technical team
receives data from dealers and schedules the entire process based on the algorithm we suggested, after which
the contents of the process are forwarded to production level for further execution.
Thus, as economies progress, the way manufacturing contributes to the economy shifts: in today's advanced
economies, manufacturing fosters creativity, competitiveness, and exports more than income and jobs.
Manufacturing in these jurisdictions has also evolved to utilize more commodities and rely more heavily on
them to function. Consider the following five items manufactured in the sector.
When a process with a higher priority enters the waiting list, priority isgranted based on the highest quantity
of goods requested, which has substantial client demand. This is preemptive because a process that is
currently runningcan be interrupted to allow a higher priority process to run. And then the technical team will
assess it and provide the product list to the manufacturing team, which will be the first to be built.
Preemptive scheduling allots time to processes for a limited duration, whereas non-preemptive scheduling
allots time to processes until they terminate or move to the waiting state. When a higher priority task comes
while a lower priority job is running, it replaces the lower priority task and then temporarily suspended until it
completes the implementation. The technical efficiency measures the ratio between source and destination
supply.
Monetary well-being is generated during the manufacturing operation, which includes all business activities
that strive to accommodate basedesires and demands wholly or partly. As a conclusion, we help to raise the
processing capability of industrial enterprises and building a good supply chain. It also allows smart industrial
processes and economic growth catalysts.
Fig.4.1. Products Manufactured Based On Priority
The product 2 has the highest demand among dealers so the product 2 willbe manufacturing first. It enables
the dealers to meet the ongoing demand by thecustomers. The level of productivity is the most fundamental
and important factor determining the standard of living. Raising it allows people to get what they want faster
or get more in the same amount of time.

49
Benefits of increased productivity:
 Creating a strong atmosphere
 Improving customer service
 Increases profitability
 Decreasing the overhead costs
 Optimizes resource use
 Promotes the organization for growth
 Improves competitiveness
 Greater customer satisfaction
4.1. System Architecture
Dealer login to his/her account once registered, and order the products from the list which is uploaded by the
admin. Once the dealers order the products he will be redirect to the payment process and then he/she may
logout. The admin approves the order done by the dealers and monitor the technical, production, storage team.
The technical team involves in analyzing which product must be manufactured based on the demand from the
dealers. After it is scheduled, the list will be sent to the production team for manufacturing. It involves three
steps which are assembling, testing and quality checking. After manufactured, the products will be sent to the
storage team.
Fig.4.2. System Architecture
5. Design and Implementation
 Dealer
 Admin
 Technical Team
 Production
 Storage

50
5.1. Dealer
In the application, the Dealer is going to register and log in to the application with username, email,
phone number and password. Then Dealer sees an available product based on demand in which the product
list is uploaded by the admin to order the products. The dealer will select and order the products with the
required quantity. Thus, Quantity will differ based on the demand, by the consumers. Then the order is placed
successfully, and the dealer will redirect to the payment option to fill in the details and to pay. Finally, dealer
will log out from the application.
Fig.5.1. Dealer flowchart
5.2. Admin
Admin logged in to the page then uploads the details of the each product detail to view by the dealer. Then
admin monitor each field in this application. Admin will check the dealer ordered list.
Fig.5.2. Admin flowchart
And the details will be sending to the technical team to analyze the list, to overcome demand. And also admin
monitor the production and storage module and check technical team who analyzed the list. The process is
updated when a huge number of dealers order the demanded products and once the process is finished admin
will log out.
5.3. Technical Team
When the dealers ordered the product, the technical team will log in and check the ordered list of the products
then the team starts the analysis of which product to manufacture first. Applying the algorithm, to solve the

51
demandissues by manufacturing the highest demanded products first. The analyzed list will be provided to the
production department as well as admin for monitor and the team will log out.
Fig.5.3. Technical flowchart
5.4. Production
When the analyzed list has arrived the production team first logged in. Then starting the production process
which is highly demanded. Initiating by assembling process followed by testing and quality checking and
both the processes is performed manually. Then the products to be produced are listed and sent to the storage
department and admin.
Fig.5.4. Production Flowchart
5.5. Storage
Then the storage department is logged in and the manufactured products are sent to the storage place. The
products are split based on the details of the demanded products ordered by the dealer. After stored, it will be
checked then the information is sent to the admin.
Fig.5.5. Storage Flowchart

52
6. Results & Discussions
Fig.6.1. Analyzing the products
Production team does the task of analyzing, and finding which product to bemanufactured first by applying
the scheduling algorithm.
Fig.6.2. Scheduling products based on priority
After analyzing by using the priority scheduling algorithm, found the list oforder of manufacturing products.
7. Conclusion
Production one of the most core elements in manufacturing and is crucial to what it means to be a
manufacturer. There will be no finalized commoditiesto sell to consumers if this activity didn't take place.
Producers should use reliable forecasting tools that can help them estimate what and where, which items to
deliver. In the coming years, since they are some prevalent difficulties in the manufacturing business to
predict customer requirements. As a result, we employed Priority scheduling algorithms for manufacturing
processes that are performed quickly and successfully, permitting to plan algorithm that has diversity

53
in different production process. By enabling scheduling algorithm manufacturing industry can increase the
production to a higher degree. Dynamicproduction makes the business to a higher level. Thus, our algorithm
helps to schedule dynamically changing process which not only makes production better also the business
growth by consumer’s satisfaction and also creating a good supply chain.
In future, we can use Artificial Intelligence related concepts for manufacturing process when the demand
arises for the new products which comes to the market, then it automatically scheduled the process for better
efficient results of the manufacturing process and also time consumption is reduced. So, it may reduce man
power and increased production level compared to our proposed method.
Declarations
Source of Funding
This research did not receive any grant from funding agencies in the public or not-for-profit sectors.
Consent for publication
Authors declare that they consented for the publication of this research work.
References
[1] Narmanov, Ulugbek. (2022). The role and importance of the digital economy in the development of
innovative. Linguistics and Culture Review, 6: 121-133.
[2] Qi, Qinglin, Fei Tao, Ying Cheng, Jiangfeng Cheng, and A. Y. C. Nee. (2021). New IT driven rapid
manufacturing for emergency response. Journal of Manufacturing Systems, 60: 928-935.
[3] Kumar, T. Ananth, S. Arunmozhi Selvi, R. S. Rajesh, and G. Glorindal. (SWI 2020). Safety Wing for
Industry–An Advanced Unmanned Aerial Vehicle Design for Safety and Security Facility Management in
Industries. In Industry 4.0 Interoperability, Analytics, Security, and Case Studies, pp. 181-198. CRC Press.
[4] Abbassi, Wyssal, Aida Harmel, Wafa Belkahla, and Helmi Ben Rejeb. (2022). Maker movement
contribution to fighting COVID‐19 pandemic: insights from Tunisian FabLabs. R&D Manag, 52(2): 343-355.
[5] Rajmohan, R., T. Ananth Kumar, M. Pavithra, S. G. Sandhya, E. G. Julie, J. J. V. Nayahi, and N. Z.
Jhanjhi. (2020). Blockchain: Next-generation technology for industry 4.0. Blockchain Technology, 177-198.
[6] Lee, In, and Kyoochun Lee. (2015). The Internet of Things (IoT): Applications, investments, and
challenges for enterprises. Business horizons, 58(4): 431-440.
[7] Doolen, Toni L., and Marla E. Hacker. (2005). A review of lean assessment in organizations: an
exploratory study of lean practices by electronics manufacturers. J. of Manufacturing systems, 24(1): 55-67.
[8] Kumar, K. Suresh, AS Radha Mani, S. Sundaresan, T. Ananth Kumar, and Y. Harold Robinson. (2021).
Blockchain-based energy-efficient smart green city in IoT environments. In Blockchain for Smart Cities, pp.
81-103. Elsevier.

54
[9] Velammal, M. Navaneetha, Tamilarasan Ananth Kumar, M. Steffi Anto, and A. Andrew Roobert. (2021).
Design of High-Speed Nanoscale Adder Logic Circuit for Low Power Consumption. In 2021 IEEE Pune
Section International Conference (PuneCon), pp. 1-6. IEEE.
[10] Lee, Jian-Yuan, Jia An, and Chee Kai Chua. (2017). Fundamentals and applications of 3D printing for
novel materials. Applied materials today, 7: 120-133.
[11] Kumar, T. Deva, TS Arun Samuel, and T. Ananth Kumar. (2021). Transforming Green Cities with IoT:
A Design Perspective. In Handbook of Green Engineering Technologies for Sustainable Smart Cities, pp.
17-35. CRC Press.
[12] Ayvaz, Serkan, and Koray Alpay. (2021). Predictive maintenance system for production lines in
manufacturing: A machine learning approach using IoT data in real-time. Expert Systems with Applications,
173: 114598.
[13] Ngo, Tran My, Tien Hoang Thuy Le, and Yen Thi Bach Tran. (2022). Innovation and firm performance:
Is R&D worth it? An empirical case of Vietnam enterprises. Science & Technology Development
Journal-Economics-Law and Management, 6(1): 2039-2050.
[14] Kumar, T. Ananth, A. John, and C. Ramesh Kumar. (2020). IoT technology and applications. Internet of
Things: From the Foundations to the Latest Frontiers in Research, 43.
[15] Gao, Wei, Qinghao Hu, Zhisheng Ye, Peng Sun, Xiaolin Wang, Yingwei Luo, Tianwei Zhang, and
Yonggang Wen. (2022). Deep Learning Workload Scheduling in GPU Datacenters: Taxonomy, Challenges
and Vision. arXiv preprint arXiv:2205.11913.
[16] Rajakumar, G., and T. Ananth Kumar. (2022). Design of Advanced Security System Using Vein Pattern
Recognition and Image Segmentation Techniques. In Advance Concepts of Image Processing and Pattern
Recognition, pp. 213-225. Springer, Singapore.
[17] Richter, Lucas, Malte Lehna, Sophie Marchand, Christoph Scholz, Alexander Dreher, Stefan Klaiber,
and Steve Lenk. (2022). Artificial Intelligence for Electricity Supply Chain automation. Renewable and
Sustainable Energy Reviews, 163: 112459.
[18] Devi, A., M. Julie Therese, P. Dharani Devi, and T. Ananth Kumar. (2021). IoT-Based Smart Pipeline
Leakage Detecting System for Petroleum Industries. In Industry 4.0 Interoperability, Analytics, Security, and
Case Studies, pp. 149-168. CRC Press.
[19] Usharani, S., P. Manju Bala, R. Rajmohan, T. Ananth Kumar, and M. Pavithra. (2022). Blockchain
Technology Use Cases in Healthcare Management: State-of-the-Art Framework and Performance Evaluation.
In Blockchain, Artificial Intelligence, and the Internet of Things, pp. 117-140. Springer, Cham.
[20] Zhao, Xiaoyang, and Chuan Wu. (2022). Large-scale Machine Learning Cluster Scheduling via
Multi-agent Graph Reinforcement Learning. IEEE Transactions on Network and Service Management.
[21] Macenski, Steven, Tully Foote, Brian Gerkey, Chris Lalancette, and William Woodall. (2022). Robot
Operating System 2: Design, architecture, and uses in the wild. Science Robotics, 7(66): eabm6074.

55
[22] Lohitha, M., E. J. Priyadharsini, K. Sangeetha, J. E. Jeyanthi, and T. Ananth Kumar. (2021). Automation
in vertical gardening using LABVIEW. In AIP Conf. Proceedings, 2378(1): 020001. AIP Publishing LLC.
[23] Paul, V. Milner, SR Boselin Prabhu, T. Jarin, and T. Ananth Kumar. (2021). Critical Investigation and
Prototype Study on Deep Brain Stimulations: An Application of Biomedical Engineering in Healthcare. In
Handbook of Deep Learning in Biomedical Engineering and Health Informatics, pp. 103-135. Apple
Academic Press.
[24] Kumar, P. Praveen, and K. U. M. A. R. T Ananth. (2018). Risk Analysis on Drilling Rig in Oil and
Natural Gas Using Engineering Tools. i-Manager's Journal on Future Engineering and Technology, 13(3): 50.
[25] Zhang, Weiwei, Guang Yang, Nan Zhang, Lei Xu, Xiaoqing Wang, Yanping Zhang, Heye Zhang, Javier
Del Ser, and Victor Hugo C. de Albuquerque. (2021). Multi-task learning with multi-view weighted fusion
attention for artery-specific calcification analysis." Information Fusion 71: 64-76.

Conflict-Receptive and Prognosis Scheduling in Deep Learning Systems

More Related Content

Similar to Conflict-Receptive and Prognosis Scheduling in Deep Learning Systems (20)

More from Associate Professor in VSB Coimbatore (20)

Recently uploaded (20)

Conflict-Receptive and Prognosis Scheduling in Deep Learning Systems