Trending 15 Artificial Intelligence Research Topics

Description of the topic

Several direction-oriented apps are used by people to find a way to reach the destination location (Cui and Shi, 2011). These applications are useful in providing an accurate path to reach the destination but fail to provide the optimal or shortest path between the source and the destined location (Kapi, Sunar and Zamri, 2020). In this research study, this problem in the existing navigation or direction-oriented applications can be discussed in detail and AI-based solutions can be developed to resolve the problem.

Research objectives

• To identify the issues in the existing navigation apps by reviewing the current literature.
• To develop an advanced AI-based algorithm for navigation apps to facilitate them with the capability of providing the shortest or the optimal path between the source and destination.

Research questions

RQ1: What are the major weaknesses of the existing navigation apps available on the web?

RQ2: How can AI-based algorithms help in enhancing the capabilities of existing navigation applications?

Research methodology

To develop the AI search algorithm for finding the shortest path, quantitative research methodology can be used in this research in which multiple search algorithms such as Breadth-first search, Depth First Search, Dijkstra and A* algorithms can be implemented (Yap, Burch, Holte and Schaeffer, 2011). The results of all the algorithms can be compared to determine the suitable algorithm for finding the shortest path.

References

• Cui, X. and Shi, H., 2011. Direction-oriented pathfinding in video games. International Journal of Artificial Intelligence & Applications, 2(4), p.1.
• Kapi, A.Y., Sunar, M.S. and Zamri, M.N., 2020. A review on informed search algorithms for video games pathfinding. International Journal, 9(3).
• Yap, P., Burch, N., Holte, R.C. and Schaeffer, J., 2011, August. Block A*: Database-driven search with applications in any-angle path-planning. In the Twenty-Fifth AAAI Conference on Artificial Intelligence.

Description of the topic

Transcribing audio into text is an advanced process that is beneficial for all users for different purposes such as for understanding making important notes in a class, for faster processing within a business organization etc. (Meredith, 2016) The currently used transcription systems are only able to transcribe the audio in a single language which is not suitable for the individuals or organizations operating in different countries (Yoshioka, et al, 2019). In this research study, this gap associated with the existing systems can be fulfilled with AI audio transcription systems that can be developed with some additional functionality of real-time transcription by accessing the microphone and the transcription in 7 different languages- English, Chinese, Arabic, French, Spanish, Hindi and Russian.

Research objectives

• To identify the key functionality of AI audio transcription systems by reviewing the existing systems.
• To develop an AI audio transcription system using Google speech-to-text API and deep learning techniques.

Research questions

RQ1: What are the potential benefits a multilingual audio transcription system can bring to the business organization operating in different markets across the globe?

Research methodology

Both qualitative and quantitative research methodologies can be used for this research where the literature-based analysis can be performed to analyze the existing AI audio transcription system and its key functionalities. Then, the quantitative research methodology with deep learning technique can be used along with the Google text-to-speech API for the development of multilingual audio transcription systems (Gowrishankar, and Bhajantri, 2016).

References

• Meredith, J., 2016. Transcribing screen-capture data: The process of developing a transcription system for multi-modal text-based data. International Journal of Social Research Methodology, 19(6), pp.663-676.
• Yoshioka, T. et al, 2019, December. Advances in online audio-visual meeting transcription. In 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU) (pp. 276-283). IEEE.
• Gowrishankar, B.S. and Bhajantri, N.U., 2016, October. An exhaustive review of automatic music transcription techniques: Survey of music transcription techniques. In 2016 International Conference on Signal Processing, Communication, Power and Embedded Systems (SCOPES) (pp. 140-152). IEEE.

Description of the topic

According to recent research, there are a total of 9000-11000 species of birds across the globe among which a large number of species are about to become extinct or have become extinct because of the changing environment, global warming and other natural factors (Rai, Sharma, Kumar and Kishor,2022). It is very difficult to identify or recognize each species of bird even for bird watchers as well because of their different types, name and features (Lasseck, 2018). Thus, a deep learning-based system can be developed in this research for the recognition of bird species based on the features extracted from their images.

Research objectives

• To investigate the key applications of image processing in different sectors.
• To develop an AI-based solution for the identification of bird species based on their features.
• To evaluate the performance of the CNN model in identifying the species of birds from their images.

Research questions

RQ: How can AI-based bird species detection algorithms be used in the identification of rare species with high accuracy?

Research methodology

For conducting this research, a quantitative research methodology can be used in which the deep learning-based Convolutional Neural Network model can be trained with a data set of three different species of birds- Painted Bunting, Eared grebe and Brewer Blackbird (Lee, Lee, Jeon and Smith, 2019). The data set can be gathered from the Kaggle data repository for this research and all the analysis steps and the results of the experimental analysis can be presented in the research study.

References

• Rai, B.K., Sharma, S., Kumar, G. and Kishor, K., 2022. Recognition of Different Bird Categories Using Image Processing. International Journal of Online & Biomedical Engineering, 18(7).
• Lasseck, M., 2018. Audio-based Bird Species Identification with Deep Convolutional Neural Networks. CLEF (working notes), 2125.
• Lee, S., Lee, M., Jeon, H. and Smith, A., 2019, April. Bird detection in an agricultural environment using image processing and neural networks. In 2019 6th International Conference on Control, Decision and Information Technologies (CoDIT) (pp. 1658-1663). IEEE.

Description of the topic

Chatbots, intelligent assistants and recommendation engines are some of the advanced AI-based solutions that are used by business organizations to assist their users in many ways such as by recommending products based on their previous purchases, answering their queries in real-time and many more (Divya, Indumathi, Ishwarya, Priyasankari and Devi 2018). In this research, all these benefits and applications of AI-based chatbots can be explored and a new and advanced AI-based virtual assistant can be developed using some advanced AI techniques such as NLP, NLTK, ISF (intelligent system Fieldwork) and speech techniques (Tamrakar and Wani 2021).

Research objectives

• To identify the application of virtual assistants and chatbots in different sectors.
• To develop the AI chatbot or visual assistant using NLTK and speech recognition.
• To evaluate the performance of the virtual assistant by providing real-time commands and queries.

Research questions

RQ: How can AI-based virtual assistants help to better resolve the queries of the customers than human assistants?

Research methodology

To conduct the research successfully, a suitable methodology is necessary and for this research, the quantitative research methodology can be suitable under which experimental analysis can be performed and the virtual assistant can be developed using four AI techniques- NLTK, NLP, ISF and Speech recognition (Batra, Yadav and Sharma, 2020). The data of multiple common commands that are asked from a virtual assistant can be gathered from reliable online sources and the Long short-term memory (LSTM) model can be trained to provide a real time response for the query.

References

• Divya, S., Indumathi, V., Ishwarya, S., Priyasankari, M. and Devi, S.K., 2018. A self-diagnosis medical chatbot using artificial intelligence. Journal of Web Development and Web Designing, 3(1), pp.1-7.
• Tamrakar, R. and Wani, N., 2021, April. Design and development of CHATBOT: A review. In Proceedings of International Conference On“Latest Trends in Civil, Mechanical and Electrical Engineering”. https://www. research gate. net/publication/351228837.
• Batra, A., Yadav, A. and Sharma, S.K., 2020. Connecting People Through Virtual Assistant on Google Assistant. In Proceedings of ICETIT 2019: Emerging Trends in Information Technology (pp. 407-417). Springer International Publishing.

Description of the topic

Several anomalies on the road are experienced by the passengers every day impacting the comfort of their rides (Bibi et al, 2021). There are several reasons for these anomalies such as low-quality construction materials, heavy vehicles and some other natural factors (Wu, et al, 2020). In this research study, the problem of road anomalies and their impact on the safety of the passengers are duly considered throughout the research and the technological solutions can be used to avoid these anomalies on the road or to reduce the impact of these anomalies by its early detection.

Research objectives

• To develop an AI-based solution for detecting road anomalies to avoid the critical problems passengers face due to these anomalies.
• To investigate the AI techniques that can be used for the detection of these anomalies and then implement them during the research to evaluate its performance.

Research questions

RQ1: How can the AI-based road anomaly detection algorithm help in enhancing the safety of passengers and drivers from road accidents due to the deep pitches on roads?

Research methodology

To find the answers to these questions and to meet the defined objectives of the research, the quantitative methodology can be followed for successfully conducting this research. Following this methodology, a Deep neural network-based model can be implemented along with the integration of an Intelligent transportation system and a Vehicular ad hoc network for real-time processing of road images and detection of road anomalies (Liu, et al., 2022).

References

• Bibi, R. et al, 2021. Edge AI-based automated detection and classification of road anomalies in VANET using deep learning. Computational intelligence and neuroscience, 2021, pp.1-16.
• Wu, C., et al, 2020. An automated machine-learning approach for road pothole detection using smartphone sensor data. Sensors, 20(19), p.5564.
• Liu, C.,et al., 2022. A response-type road anomaly detection and evaluation method for steady driving of automated vehicles. IEEE Transactions on Intelligent Transportation Systems, 23(11), pp.21984-21995.

Description of the topic

The use of artificial intelligence is increasing in almost every sector considering the advancements and benefits it has brought for individuals, business organizations and the government authorities as well (Thuraisingham, 2020). Cybersecurity is another sector where this technology is increasingly used for automated authentication, image processing, network security operations etc. (Shamiulla, 2019). There are several achievements of AI in this field that are investigated in this research to identify the impact of this technology on the increased number of cyber attacks.

Research objectives

• To identify the achievements of AI in the field of cyber security from literature-based analysis.
• To determine the positive and negative aspects of AI in cyber security from the perspectives of cyber security experts.

Research questions

RQ1: To what extent, AI technology has impacted the field of cyber security with its advanced solutions?

Research methodology

A mixed approach can be followed in this research study where the literature-based analysis and one-on-one interviews with the cyber security experts can be conducted to gather their responses regarding the positive and negative aspects of AI in cyber security. Also, a survey-based analysis can be performed to determine the impact of AI on the number of cyber attacks and technology exploitation in an organization (Das, Balmiki and Mazumdar, 2022). Participants for the survey can be UK-based organizations who have been using these technologies for strengthening their security measures.

References

• Thuraisingham, B., 2020, May. The role of artificial intelligence and cyber security for social media. In 2020 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW) (pp. 1-3). IEEE.
• Shamiulla, A.M., 2019. Role of artificial intelligence in cyber security. International Journal of Innovative Technology and Exploring Engineering, 9(1), pp.4628-4630.
• Das, S., Balmiki, A.K. and Mazumdar, K., 2022. The Role of AI-ML Techniques in Cyber Security. In Methods, Implementation, and Application of Cyber Security Intelligence and Analytics (pp. 35-51). IGI Global.

Description of the topic

Defective products or missing elements in a product are some of the common issues that are faced by manufacturers during production. During medicine packaging there are several cases of defects such as opened packets, missing medicines, not satisfying the quality standards etc. (Karthikeyan, Tiwari, Zhong and Bukkapatnam, 2022) The quality monitoring systems used in the manufacturing sites only monitor the quality of the medicines while the packaging of the medicines are often neglected (Sheu, Chen, Pardeshi, Pai and Chen, 2021). This not only disrupts the manufacturing KPIs of the healthcare sector but also negatively contributes to customer dissatisfaction. To resolve this problem, an AI-based automated defect detection system can be developed during this research which can be scanning the product and can process the elements of the package in real-time to identify any defects.

Research objectives

• To investigate the negative impacts of defects in the medicine packaging on the quality of medicine and the manufacturers.
• To develop an AI-based defect detection system for automated defect detection with real-time processing of the scanned image of the product.
• To evaluate the performance of the AI-based solution based on several metrics.

Research questions

RQ: How can advanced technologies such as AI and Deep learning help in enhancing the quality of medicine packaging?

Research methodology

In this research, quantitative research methodology can be used in which an experimental analysis process can be performed for the development of a deep learning-based model to detect the defects in the medicine packaging (Chouchene et al., 2020). The model can be trained with the secondary data of the images of high-quality medicine packaging and the images of defective medicine packaging.

References

• Karthikeyan, A., Tiwari, A., Zhong, Y. and Bukkapatnam, S.T., 2022. Explainable AI-infused ultrasonic inspection for internal defect detection. CIRP Annals, 71(1), pp.449-452.
• Sheu, R.K., Chen, L.C., Pardeshi, M.S., Pai, K.C. and Chen, C.Y., 2021. AI Landing for Sheet Metal-Based Drawer Box Defect Detection Using Deep Learning (ALDB-DL). Processes, 9(5), p.768.
• Chouchene, A., et al., 2020, February. Artificial intelligence for product quality inspection toward smart industries: quality control of vehicle non-conformities. In 2020 9th international conference on industrial technology and management (ICITM) (pp. 127-131). IEEE.

Description of the topic

The constraint satisfaction problems include scenarios where some possible set of values or conditions are specified that must be maintained while evaluating the solution (Mehta, 2021). It is majorly used in games such as crossword puzzles, sudoku etc. It is very difficult to find the appropriate solutions for these problems used in the games such as Sudoku where the values need to be filled in the empty boxes (Maji and Pal, 2014). To evaluate the values to be filled in the empty boxes in a sudoku board, a reinforcement learning algorithm can be developed in this research.

Research objectives

• To identify the reinforcement learning techniques that can be used for resolving constraint satisfaction problems.
• To develop an AI-based backtracking algorithm for solving the sudoku puzzle.
• To test the performance of the backtracking algorithm in evaluating the missing values on the sudoku board.

Research questions

RQ: How does the backtracking algorithm help in accurately evaluating the missing values on the sudoku board?

Research methodology

This research can be conducted using the quantitative research methodology in which experimental analysis can be performed to use the backtracking algorithm for determining the empty blocks on the sudoku board and then to find the values for those empty blocks considering the specified constraints (Schottlender, 2014).

References

• Mehta, A., 2021. Reinforcement learning for constraint satisfaction game agents (15-puzzle, minesweeper, 2048, and sudoku). arXiv preprint arXiv:2102.06019.
• Maji, A.K. and Pal, R.K., 2014, February. Sudoku solver using mini grid-based backtracking. In 2014 IEEE International Advance Computing Conference (IACC) (pp. 36-44). IEEE.
• Schottlender, M., 2014, May. The effect of guess choices on the efficiency of a backtracking algorithm in a Sudoku solver. In IEEE Long Island Systems, Applications and Technology (LISAT) Conference 2014 (pp. 1-6). IEEE.

Description of the topic

Modern technologies and machines have revolutionized the agriculture sector in many ways such as sensors for monitoring crops, automated sprinklers, advanced machinery for harvesting etc. (Orchi, Sadik and Khaldoun 2022) But even with these advanced systems, the identification of crops' situation, need of harvesting and other characteristics is a difficult task with changing climatic conditions and thus the change in their evolution cycle (Chougule and Mashalkar, 2022). In this research, a suitable solution for this problem can be provided with the development of an AI-based crop detection system that can estimate the crop density, height, type etc. using radar images to determine different needs of a crop in different situations (Efremova West and Zausaev, 2019).

Research objectives

• To study the different characteristics of crops and their evolution cycle.
• To build a deep learning-based model using the data of different types of crop images.
• To test the real-time performance of the model in determining the characteristics of the crop.

Research questions

RQ: How do deep learning techniques help in enhancing crop detection for horticulture and agriculture specialists?

Research methodology

To address the defined objectives of the research, a quantitative methodology can be used. Under this methodology, an experimental analysis can be performed in which a deep learning model can be built using the data of radar images of crops gathered from reliable online resources. In addition to that, a literature-based analysis can also be performed to find effective models for this purpose by reviewing the existing projects.

References

• Orchi, H., Sadik, M. and Khaldoun, M., 2022. On using artificial intelligence and the internet of things for crop disease detection: A contemporary survey. Agriculture, 12(1), p.9.
• Chougule, M.A. and Mashalkar, A.S., 2022. A comprehensive review of agriculture irrigation using artificial intelligence for crop production. Computational Intelligence in Manufacturing, pp.187-200.
• Efremova, N., West, D. and Zausaev, D., 2019. AI-based evaluation of the SDGs: the case of crop detection with earth observation data. arXiv preprint arXiv:1907.02813.

Description of the topic

The plants are prone to several diseases because of changes in climatic conditions, pests, nutritional deficiency etc. (Sardogan, Tuncer, and Ozen, 2018) Similarly, for a tomato plant these factors can lead to the emergence of several dangerous diseases which can impact the whole crop yield (Annabel, Annapoorani and Deepalakshmi, 2019). These diseases can be identified from the change in the leaves of the plant (Trivedi, Shamnani, and Gajjar, 2020). In this research, an advanced method can be introduced to provide an automated solution for the detection of diseases in a tomato plant.

Research objectives

• To identify the factors that cause the diseases in the tomato plants along with the details of some common diseases in tomato plants.
• To provide an automated solution for the detection of diseases in a tomato plant.
• To evaluate the performance of the automated solution to ensure its reliability and accuracy.

Research questions

RQ: How does the AI algorithm contribute to the enhancement of the detection and diagnosis of plant diseases?

Research methodology

This research can be conducted using the quantitative methodology in which the experimental analysis can be performed to meet the proposed objectives. Deep learning techniques can be used for building the system to detect plant leaf diseases. The secondary data of the leaf images with three different diseases can be gathered from the reliable online data set repository Kaggle.

References

• Erdogan, M., Tuncer, A. and Ozen, Y., 2018, September. Plant leaf disease detection and classification based on CNN with LVQ algorithm. In 2018 3rd international conference on computer science and engineering (UBMK) (pp. 382-385). IEEE.
• Annabel, L.S.P., Annapoorani, T. and Deepalakshmi, P., 2019, April. Machine learning for plant leaf disease detection and classification–a review. In 2019 international conference on communication and signal processing (ICCSP) (pp. 0538-0542). IEEE.
• Trivedi, J., Shamnani, Y. and Gajjar, R., 2020. Plant leaf disease detection using machine learning. In Emerging Technology Trends in Electronics, Communication and Networking: Third International Conference, ET2ECN 2020, Surat, India, February 7–8, 2020, Revised Selected Papers 3 (pp. 267-276). Springer Singapore.

Description of the topic

Overspeeding, driver drowsiness, road anomalies etc. are some of the reasons that lead to road accidents (Liao et al, 2020). These issues have impacted the safety of the drivers and other passengers. With the lack of limiting controls and automated operating, the cars are not prevented from getting out of control and this increases the chances of road accidents (Rasheed, Hu and Zhang, 2020). To address this issue, a deep reinforcement learning algorithm can be proposed in this research to enable some automated operations such as automated movement of the steering wheel, limiting speed etc.

Research objectives

• To investigate the major reasons for the increasing road accidents.
• To discuss the application of deep reinforcement learning in building automated solutions.
• To build an AI algorithm for enabling self-driving capabilities in cars using multiple reinforcement learning techniques.

Research questions

RQ: How can AI-based algorithms help in reducing the number of road accidents?

Research methodology

Quantitative research methodology can be used for conducting this research in which deep reinforcement and inverse reinforcement learning techniques such as Q learning, and A* algorithms can be used to enable self-driving capabilities in modern cars (Kiran, et al., 2021). The legal policies for autonomous cars and the use of AI in manufacturing can be considered while conducting this research.

References

• Liao, J et al, 2020. Decision-making strategy on the highway for autonomous vehicles using deep reinforcement learning. IEEE Access, 8, pp.177804-177814.
• Rasheed, I., Hu, F. and Zhang, L., 2020. Deep reinforcement learning approach for autonomous vehicle systems for maintaining security and safety using LSTM-GAN. Vehicular Communications, 26, p.100266.
• Kiran, B.R.,et al., 2021. Deep reinforcement learning for autonomous driving: A survey. IEEE Transactions on Intelligent Transportation Systems, 23(6), pp.4909-4926.

Description of the topic

Artificial Intelligence has transformed the whole world with automated systems and technological solutions for every problem (Makhni, Makhni and Ramkumar, 2021). The positive aspects of technology have overshadowed the destructive side of technology. Most of the existing research studies emphasize the benefits the technology has brought to individuals, organizations and government authorities (Prianto, Sumantri and Sasmita, 2020). But, the negative aspects of the technology are not yet explored properly in any of the research studies because of which the demerits of the technology have not been identified (Moshayedi, Roy, Kolahdooz and Shuxin, 2022). In this research, both the negative and positive aspects of the technology can be investigated considering the opinions of the business organizations who are using the technology and also the AI experts working in the field for a long time.

Research objectives

• To identify both positive and negative aspects of AI for society through literature-based analysis.
• To evaluate the impact of AI technology on society based on its positive and negative aspects.

Research questions

RQ1: What changes did AI technology bring to the world in terms of automation of daily tasks?

RQ2: Does AI have a destructive or constructive impact on Society?

Research methodology

To analyze the impact of AI on society, a qualitative research methodology can be used in which interviews of CEOs of the business organizations that have been using AI in their organization can be conducted to know their opinions about the impact of AI on society. In addition to this, online surveys can be conducted separately for the general public and for the AI experts to know their opinions as well. The responses of these participants can be gathered and the impact of AI on society can be determined.

References

• Makhni, E.C., Makhni, S. and Ramkumar, P.N., 2021. Artificial intelligence for the orthopedic surgeon: an overview of potential benefits, limitations, and clinical applications. JAAOS-Journal of the American Academy of Orthopaedic Surgeons, 29(6), pp.235-243.
• Prianto, Y., Sumantri, V.K. and Sasmita, P.Y., 2020, May. Pros and cons of AI robots as a legal subject. In Tarumanagara International Conference on the Applications of Social Sciences and Humanities (TICASH 2019) (pp. 380-387). Atlantis Press.
• Moshayedi, A.J., Roy, A.S., Kolahdooz, A. and Shuxin, Y., 2022. Deep Learning Application Pros And Cons Over Algorithm Deep Learning Application Pros And Cons Over Algorithm. EAI Endorsed Transactions on AI and Robotics, 1(1).

Description of the topic

The manufacturing of self-driving cars has become the latest trend or the latest challenge for big companies such as Tesla, Mercedes Benz, Toyota etc. (Qin and Yan 2021) But, one of the most important functions of these autonomous cars is the capability of understanding the traffic signals (Lim, Hong, Choi and Byun 2017). There are different types of traffic signs such as for speed limits, directions, for children crossing, no entry, no parking etc. the lack of this functionality in autonomous cars can lead to dangerous accidents on several occasions such as when the speed limit needs to be maintained, stoppage on traffic signals etc. (Ayachi, Afif, Said and Atri 2020) Considering this problem, a deep learning based model can be developed in this research that can enable these cars to understand all the traffic signs accurately.

Research objectives

• To identify the need for traffic sign detection functionality in autonomous cars.
• To provide an advanced solution for enabling the traffic sign detection capability in self-driving cars.
• To test the working of the algorithm to ensure its accuracy and reliability.

Research questions

RQ: How AI-deep learning techniques can enhance the capabilities of self-driving cars by improving traffic sign detection?

Research methodology

This research will be conducted to find a suitable solution for enhancing the capabilities of self-driving cars and can be conducted using quantitative research methodology. Under this methodology, a deep learning model can be built using the data of all the traffic signs to enable the traffic sign detection capabilities of self-driving cars.

References

• Qin, Z. and Yan, W.Q., 2021. Traffic-sign recognition using deep learning. In Geometry and Vision: First International Symposium, ISGV 2021, Auckland, New Zealand, January 28-29, 2021, Revised Selected Papers 1 (pp. 13-25). Springer International Publishing.
• Lim, K., Hong, Y., Choi, Y. and Byun, H., 2017. Real-time traffic sign recognition based on a general-purpose GPU and deep learning. PLoS one, 12(3), p.e0173317.
• Ayachi, R., Afif, M., Said, Y. and Atri, M., 2020. Traffic signs detection for real-world application of an advanced driving assistance system using deep learning. Neural Processing Letters, 51, pp.837-851.

Description of the topic

Handwritten digit recognition refers to the capability of the computer system to recognise the digits written by the user (Ahlawat, Choudhary, Nayyar, Singh and Yoon, 2020). The existing algorithms used for recognising human handwriting in computer systems are not reliable because of less training data and there are high chances of inaccurate recognition because of the different types of handwriting of different individuals (Niu and Suen, 2012). The issue of reliability of these algorithms can be considered in this research study and a suitable method to enhance the capability of the machines to recognize the handwritten digits can be provided.

Research objectives

• To identify the limitations of the existing handwriting recognition algorithms.
• To implement a CNN model for the accurate recognition of handwritten digits.
• To evaluate the performance of the CNN model to ensure its reliability.

Research questions

RQ: How can the AI-based algorithm be used for enhancing the capability of the machine to recognise handwriting text?

Research methodology

This research can be conducted using the quantitative methodology in which the experimental analysis process can be performed using the data of 50,000 images of digits from 0 to 9 written by different individuals. CNN models can be trained with the data of 50,000 images for the accurate recognition of handwritten digits.

References

• Ahlawat, S., Choudhary, A., Nayyar, A., Singh, S. and Yoon, B., 2020. Improved handwritten digit recognition using convolutional neural networks (CNN). Sensors, 20(12), p.3344.
• Niu, X.X. and Suen, C.Y., 2012. A novel hybrid CNN–SVM classifier for recognizing handwritten digits. Pattern Recognition, 45(4), pp.1318-1325.

Description of the topic

Physical security is also an important aspect of an IT organization along with online security measures as the physical assets of the company are also equally critical for the organization (Kure, Islam and Razzaque, 2018). Most organizations do not focus more on their physical security and thus, increase the chances of social engineering attacks or insider attacks (Zafar, 2013). To address these privacy issues, an AI-based security system is proposed in this research which can be trained to recognize the faces of the company’s employees and the access rights so that only the authorized member of the company is allowed to access the particular spaces within the office (Li, Peng, Qiao and Peng, 2019).

Research objectives

• To enhance the physical security measures of the IT organization.
• To perform an experimental analysis to build a deep learning-based image recognition system.
• To evaluate the accuracy of the built deep learning model to ensure its high-quality performance.

Research questions

RQ: How can deep learning-based AI solutions enhance the physical security of IT organizations?

Research methodology

This research can be conducted using the quantitative research methodology in which the experimental analysis can be performed using the data set of the images of the employees of the IT organization. Access rights can be provided to the employees of the company to build the CNN model for accurate image recognition while authenticating the identity of the employees.

References

• Li, Q., Peng, X., Qiao, Y. and Peng, Q., 2019. Learning category correlations for multi-label image recognition with graph networks. arXiv preprint arXiv:1909.13005.
• Zafar, H., 2013. Human resource information systems: Information security concerns for organizations. Human Resource Management Review, 23(1), pp.105-113.
• Kure, H.I., Islam, S. and Razzaque, M.A., 2018. An integrated cyber security risk management approach for a cyber-physical system. Applied Sciences, 8(6), p.898.