Change detection is widely used to estimate deforestation, LULC, urban growth, shifting cultivation, etc. If the change detection process on earth surface is carried out accurately and timely, then the interaction and relationship between humans and natural phenomena can be better investigated leading to a better utilization of the natural resources.

Read the following articles to get an idea about the topic –

i) https://www.aclweb.org/anthology/2020.lrec-1.531.pdf

ii) https://paperswithcode.com/task/abuse-detection

iii) http://sersc.org/journals/index.php/IJAST/article/view/14949

Please go through following articles and discuss the problem statement with me (Deepak Uniyal) before finalizing the statement. The links given below are for reference purpose and you should not exactly copy these ideas.

i) https://iopscience.iop.org/article/10.1088/1757-899X/263/4/042027/pdf

ii) https://www.instructables.com/Smart-Garbage-Monitoring-System-Using-Internet-of-/

iii) https://www.electronicshub.org/smart-dustbin-using-arduino/

Smart cities use Internet of Things (IoT) devices such as connected sensors, lights, and meters to collect and analyze data. The cities then use this data to improve infrastructure, public utilities and services, and more.

Machine learning generally takes the data generated by several apps such as Health MD applications, internet-enabled cars, etc. and leverages it to identify patterns and learn how to optimize the given set of services. Its tools are able to personalize the smart city experience by aggregating information about the most used roads in a city and then apply it to a transportation system.

On the other hand, machine learning and AI can be helpful in waste collection and its proper management and disposal which is a vital municipal activity in a city. Thus, the technology for smart recycling and waste management provides a sustainable waste management system. AI has the ability to understand how cities are being used and how they are functioning. It assists city planners in comprehending how the city is responding to various changes and initiatives.

Students can design AI enabled authentication schemes for various computing domains like smart homes, smart transportation system, smart healthcare systems.

Students can design access control schemes for Industry 4.0 communication.

Segment the customers based on their gender, age, interest. It is useful in customized marketing. Customer segmentation is an important practice of dividing customers based on individual groups that are similar.

Contrast different available tools and techniques in big data stack and analyze performance of each tool , handson on different programming techniques using real world use cases.

Machine learning has virtually endless applications in the healthcare industry. Today, machine learning is helping to streamline administrative processes in hospitals, map and treat infectious diseases and personalize medical treatments.

The main benefit of this system is that the use of keyboard is completely eliminated, the user will have to respond through voice and mouse click only.

Visual Question Answering is a research area about building an AI system to answer questions presented in a natural language about an image.

A system that solves this task demonstrates a more general understanding of images: it must be able to answer completely different questions about an image, oftentimes even addressing different sections of the image.

Twitter boasts 330 million monthly active users, which allows businesses to reach a broad audience and connect with customers without intermediaries. On the downside, there is so much information that it’s hard for brands to quickly detect negative social mentions that could harm their business.

That’s why sentiment analysis, which involves monitoring emotions in conversations on social media platforms, has become a key strategy in social media marketing.

Listening to how customers feel on Twitter allows companies to understand their audience, keep on top of what’s being said about their brand, and their competitors, and discover new trends in the industry.

Tools: MATLAB/Python, Description: An image fusion based on multimodal medical images renders a considerable enhancement in the quality of fused images. An effective image fusion technique produces output images by preserving all the viable and prominent information gathered from the source images without any introduction of flaws or unnecessary distortions.

Tool: MATLAB/Python, Description: medical imaging is frequently justified in the follow up of a disease which is already diagnosed and treated. Medical images like any other form of imaging techniques are susceptible to noise and artifacts. Noise can be random or white noise with an even frequency distribution or frequency dependent noise introduced by a device’s mechanism or signal processing algorithms. The presence of noise makes the images unclear and may perplex the identification and analysis of diseases which may result heavy losses including deaths. Hence, denoising of medical images is a mandatory and essential pre-processing technique for further medical image processing stages.

This project will allow you to control the devices and appliances present at your home via an app or even via voice commands..

Write a script that performs the following function s(features extracted by static analysis)

a. For the apks present in dataset, diassemble the apk

b. Extract the permission features from the manifest files.

c. The features extracted should be converted into a csv file. Thus for each apk there should be a row in the csv. The columns of the csv will be the different permission. The value for each apk will be 0 or 1 based on permission present or absent.

* Use a cloud platform for performing project. Dataset- Use any dataset available at this link https://github.com/traceflight/Android-Malware-Datasets

Write a script that performs the following functions (features extracted by dynamic analysis)

a. For the apks present in dataset, Runs the apk in a secure environment (in a VM)

b. Using the tools extract the API calls (AndroidTamer->MobSF can be used here) that the apk makes during its execution.

c. Converts the obtained API calls details into a csv file. Thus for each apk there should be a row in the csv. The columns of the csv will be the different API calls. The value for each apk will be 0 or 1 based on API call present or absent

Android Tamer is a VM consisting of several Android analysis tools ranging from analysis to pentesting. 8GB RAM is must otherwise the speed of the application slows down w.r.t. responsiveness.

Perform penetration testing of the web based mobile apps (use the BurpSuite for this). Exploit the XSS vulnerability by using browser on the emulator. You will need to set up proxy setting on the emulator. Setting will also be made in the Burpsuite to intercept the requests from app and then these requests can be tampered. Create a demo app(web based) and include a search text box in this. When you query using this search box, you can intercept this search string using Burp and then also tamper it.

Try to exploit the vulnerable components of an app (benign app) from another app (malicious app). Do the following

a. Create a demo app with a login screen

b. Once valid credentials are provided it should take you to the next screen/page.

c. Try to open the second screen/page directly from the malicious app.

d. Then also provide a fix for this vulnerability.

e. Once done, now again try to access the second screen/page from the malicious app by encoding permissions in the malicious app.

This scenario demonstrates how a malicious app tries to access secure area of a genuine app with code vulnerabilities.

Demonstrate and compare the performance of any three web browser forensics tools.  Analysis the activity logs of any one web browser.

Find the vulnerabilities in any IoT device using penetration testing tools. Parse the logs and generate some report as an output.

Develop and host some innovative web application using cloud services such as AWS EC2, AWS S3 and RDS services etc.

Extract the logs of services associated with your account. Parse them and convert into some meaningful format using python APIs. You can use CloudTrail, CloudWatch  etc. to monitor the activity logs. Write a script to extract and parse such logs.

Purpose:  To compare and evaluate the performance of different tools by using quantifiable data. To find out the best tool after running test cases on 4-5 different forensic tools.

Questions to be answered:

i)The tool’s features: How does it process, index, search, etc.?

ii)The tool’s ability to obtain artifacts:  How many artifacts?,  Location of those artifacts, How many relevant artifacts? , How quickly were artifacts obtained? (time)

iii)The tools searching speed: What differences are present between the tool’s searches?, Test under different scenarios

iv)The tool’s ability to recover deleted files: What data can be recovered?, Can it be used to create a time series analysis with support of artifacts?

v)The tool’s level of difficulty:  Is it easy for beginners?, Designing test cases is easy?, Evaluate the analysis provided by the tool.

Testbed setup:

Hardware used: Use virtual machines for testing. Do not use your personal system as there are high chances of damage. Use a virtual machine via VMware or VMWorkstation. Run the test case scenario to setup the scene. Create a copy of the VM by using the tool FTK Imager.

Software used:

i)FTK Imager

ii)FTK 6.1 (Forensic Toolkit)

iii)SANS Investigative Forensic Toolkit Workstation

iv)EnCase Forensic v8

v)Magnet IEF

vi)VMware vSphere

JPEG (other formats) – analysis of image formats in order to find out hidden information which might be stored. It can be extended by doing analysis of various other multimedia files including audio, video, network packets. A detailed study of formats along with their threat vectors has to be done.

Change detection is widely used to estimate deforestation, LULC, urban growth, shifting cultivation, etc. If the change detection process on earth surface is carried out accurately and timely, then the interaction and relationship between humans and natural phenomena can be better investigated leading to a better utilization of the natural resources.

Smart cities use Internet of Things (IoT) devices such as connected sensors, lights, and meters to collect and analyze data. The cities then use this data to improve infrastructure, public utilities and services, and more.

Machine learning generally takes the data generated by several apps such as Health MD applications, internet-enabled cars, etc. and leverages it to identify patterns and learn how to optimize the given set of services. Its tools are able to personalize the smart city experience by aggregating information about the most used roads in a city and then apply it to a transportation system.

On the other hand, machine learning and AI can be helpful in waste collection and its proper management and disposal which is a vital municipal activity in a city. Thus, the technology for smart recycling and waste management provides a sustainable waste management system. AI has the ability to understand how cities are being used and how they are functioning. It assists city planners in comprehending how the city is responding to various changes and initiatives.

Students can design some scheme for secure machine learning models.

Students can design AI enabled blockchain based security techniques for various computing domains like smart homes, smart transportation system, smart healthcare systems.

Audio- and video-based emotion recognition in the wild is still a challenging task. In the wild (as opposed to controlled) refers to real-life and uncontrolled conditions of data acquisition, including changing and challenging lighting conditions, indoor and outdoor scenarios, sensor and environmental noise and resolution issues, motion blur, occlusions and pose changes, as well as uncontrolled scenarios from which the emotional expressions are produced.

Incorporate Data Mining in Big Data World Real Time applications, Work on Structure and Graph Mining, Analysis on relationship between different attributes using graph mining.

In today’s competitive healthcare industry, hospitals, physician practices, and health systems are becoming more focused on improving the patient experience. Healthcare facilities offering convenience, functionality, and access that modern consumers expect are emerging leaders in their spaces and improving their bottom-line.

Today’s consumers have a ton of options when it comes to choosing healthcare services. Therefore, healthcare facilities need to be proactive in attracting and retaining patients. One of the ways they can do this is by using patient sentiment analysis.

Hate Speech Detection is the automated task of detecting if a piece of text contains hate speech.

Tool: Matlab/ Python, Description:  In this project, two different modality medical images are fused and denoised using deep learning concept to get single more informative output medical fused image.

Tools: Matlab/Python; Description: The detection, segmentation, and extraction from Magnetic Resonance Imaging (MRI) images of contaminated tumor areas are significant concerns; however, a repetitive and extensive task executed by radiologists or clinical experts relies on their expertise. Image processing concepts can imagine the various anatomical structure of the human organ. Detection of human brain abnormal structures by basic imaging techniques is challenging.

This system will automate complete traffic signaling system and will monitor traffic signal densities. This system will provide an option to the controllers to override any signal in case of emergency.

Investigate the type of Malware, e.g. Botnets in the virtualized environment. Detect the malware infection, propagation and/or attack in the virtualize environment and then in cloud computing.

Machine learning is a popular approach to malware detection because it can generalize to never-before-seen malware families and polymorphic strains.  Recent work in adversarial machine learning has shown that models are susceptible to gradient-based and other attacks. So, to study/investigate the various invasion techniques against the state-of-the-art malware detection techniques.

The various Machine Learning techniques have been applied to Malware Detection. So, to study the state-of-the-art machine learning techniques used for Malware detection and develop a model to detect the latest robust and stealthy malware in the wild.

Sentiment analysis is a kind of data mining where you measure the inclination of people’s opinions by using NLP, text analysis, and computational linguistics. We perform sentiment analysis mostly on public reviews, social media platforms, and similar sites.

Recommendation System is a filtration program whose prime goal is to predict the rating or preference of a user towards a domain-specific item or item.

Write a script to perform the following functionality.

a)         Using Convolution neural network, detect the IoT malware as dynamic analysis. (Requirements: VM, Raspberry pi 3B+ model, Cloud Environment).

b)         Analysis of cyber attack for Raspberry pi Malware.

(Requirement: AWS Cowrie services, Ubuntu Server, Mirai code (download from github), Raspberry Pi).

Write a script to generate and analyze DDoS attack on the Raspberry pi model integrated with AWS IoT Core

Mirai Dataset analysis using Cloud based Machine Learning Platform.

The IDS system should be safe from the machine learning limitations which sometimes cause a system to prone to evasion and poisoning attacks.

Basically, you have to work on designing secure ML based algorithm for android malware analysis. The dataset info will be shared by the Resource person

Design and implement an efficient web browser forensics system. Consider any web browser such as Google Chrome or Firefox for analysis.

Demonstrate the attack scenario using HInjector (tool that has a series of hypercall attacks) that can be used for testing that can be used to exploit the hypervisor (Xen or KVM etc).Generate some logs.

Perform the DoS attack to exploit the hypercall interface. The virtualization platform specifications are the XenServer (latest version) and the client (XenCenter). The both guest OSs can be Ubuntu or Windows. Do the following:

a.       Create a test set up with Xen (or any other hypervisor) base virtualized platform.

b.      Create two guest virtual machines, one can be a legitimate web server and other can play the role of the attackers.

Demonstrate the impact of the virtual machine introspection tools on the performance of the virtual machines. Do the following:

a.       Install VMI tools (for example, LibVMI (library for Xen or KVM)) in Dom 0 or privileged domain.

b.      Check the performance of virtual machines while monitored.

Monitoring the hypercalls (from virtual machine to hypervisor) and collect the hypercall traces. For Xen hypervisor, a tool like Xentracer and XenAnalyzer can be used. Do the following:

a.       Hypercalls can be monitored in Dom 0 (privileged domain) using Xentrace plugin.

b.      Store them in hard disk, then XenAnalyze plugin is used to convert this data in the human readable format.

Perform the security analysis of the cloud services to find if there is any vulnerability present. You can use CloudTrail, CloudWatch  and Sagemaker etc. to monitor and analyze the activity logs of various services associated with your account. Write a script to extract, parse and analyze such logs using machine learning.

Perform and analysis the cloud based data analytic tools and algorithms to analysis the network attack dataset. You have to write the code using the cloud based ML APIs. No GUI based platform is allowed.

Genetic algorithm, ant colony optimization etc. can be used to design a model which uses an algorithm to schedule tasks – with focus on increasing performance, reducing energy consumption, reducing resource allocation time – geographically over cloud end points. Use of CloudSim simulator is encouraged as it helps to simulate the exact environment which is needed as defined by our constraints of IoT based scenario.

Implement bloom filter on cloud’s agent based or log-based data. This increases the security and trust of users as well as investigators by taking out the scope of manipulation out of the hands of CSP. Two different implementations of bloom filter can be implemented, and the results compared. Use of CloudSim simulator along with Java is recommended.

There are ways to get hold/ownership of a virtual machine in a cloud environment. In this project, we aim to demonstrate the exploitation of an unprotected virtual environment and acquire virtual machine data. All the ways through which this can be done will be exhaustively covered one by one by attacking the VM’s through different means. Any established penetration testing software tool will be used (can use multiple) along with VMware Virtual Infrastructure. By working on this, we will generate a comprehensive guide as to how a VM can be exploited and a comparative analysis of “security of traditional hardware” as compared to “virtual hardware”.