My AI Blogging Internship with The Research Nest
And how they inspired me to pursue projects afterwards!
Just like everyone else, the lockdown had pushed me to explore something new. I was looking for some new avenues to improve my personal portfolio and so I applied to blog with TRN (The Research Nest). My main topic of focus was on Artificial Intelligence and I delivered 10 articles during the tenure. Here are my major takeaways from the experience —
- The feedback received from Aditya (Founder of TRN) helped me understand where I lacked on technical aspects and how to keep readers engaged throughout the content.
- I was complete unaware about the scope of artificial intelligence in general. This helped me realize the potential of AI and, approach and explore cross domain projects.
- I also came across a few writing tools to help me strengthen the quality of the articles.
Now I will list down short summaries of the 10 articles which I worked on during the internship and how I proceeded on a project to extend on it.
Explainable AI — What Is It And Why Do We Need It?
Here I introduce the term Explainable AI and help someone dive intro the domain. I then continue to explain the demand of the field and how state of the art solutions now require it as a stamp of guarantee to be deployed in the real world. The blog continues to explain recent advancements such as AI 360 by IBM and Google’s What If tool. The blog ends on a final discussion and a future standpoint of the domain. I personally contributed to XAI post this through this research preprint: SS-CAM: Smoothed Score-CAM for Sharper Visual Feature Localization.
Generating art with AI-powered applications
![𝒎𝒊𝒏 𝑮 𝒎𝒂𝒙 𝑫 𝔼𝒙 [𝒍𝒐𝒈 𝑫 (𝒙))] + 𝔼𝒛 [𝒍𝒐𝒈(𝟏 − 𝑫(𝑮(𝒛)))], Portrait of Edmond de Belamy, from La Famille de](https://miro.medium.com/v2/resize:fit:1400/format:webp/1*1INLfx1AV0V67lpuRgvarA.jpeg)
Creating art with AI is really an exciting domain. In 2018 an art piece generated by an AI-based algorithm was sold for about half a million dollars. I focused on introducing and explaining various software’s such as Playform, GauGAN etc. and presented my personally curated works with them. I also explore other use-cases with GANs and potential datasets which could be used along for it. I carried this excitement into my course’s minor project for extending datasets with GANs.
The Ethics of AI. What should it do in these situations?
Here we explore the computational variant of ethical laws which humans abide to. The blog explores different approaches to various existing ethical problems and, the pros and cons involved with a computer/AI taking the final decision instead of a human being. I end it on a personal note with explaining how the domain can evolve and which direction it should pursue. I took this a step further by being a part of the project on developing a privacy preserving face recognition model led by Rakshit Naidu (pre-print to come out in mid 2021).
Fun Artificial Intelligence Powered Video Games And Simulations You Should Try Now
With the diversity of AI, it was inevitable that its methods would be adopted in the domain of gaming. This blog introduces several free online games based on reinforcement learning and a short summary on the potential tech stack one should proceed with before diving into the domain. I didn’t proceed with any project here due to the computational constraints I faced due to covid.
The Evolution Of Conversational AI
Conversation is a human necessity and a skill required for any tasks. It was only a matter of time before attempts were made to automate it. This blog explores the timeline of chatbots and how they developed over the years. I went over models from core NLP to models containing 2.6 billion parameters. I recently started working on developing a chatbot focused on awareness in global warming and environmental education in general.
Top security threats in using machine learning models
Machine learning algorithms can be attacked, hacked for several reasons, and in several different ways. Here we discover whether your ML algorithms are secure and how they can be attacked. My project on preserving priavacy of models comes inline with this. Other than that, a few members of my research group are working on Audio classification against adversarial attacks.
AI that can generate music
Music drives people. The industry can be quite tasking and producing music which is loved by a greater population is quite challenging. This blog focuses on a compilation of curated efforts in the form of interesting tools and frameworks to automate the music generation process. The final part of the blog paves ways for future aspects of the domain. I proceeded with a small project on incorporating different genre’s of music to create new music using GANs.
Interesting Novels Written By Artificial Intelligence
In this blog I cover 5 created by or with the help of Machine Learning that might just bring up fear in authors as there is a book here which placed further in a competition that it’s human competitors. While the books covered are quite vague they present a great stand-point for future directions. I didn’t delve in this domain for a project but hope to in the future with the help of OpenAI’s GPT-3 Model.
Top Frameworks To Explore Reinforcement Learning
Reinforcement learning is a sub-field of machine learning which involves teaching an agent on how to take actions on the basis of a reward system. This blog covers the recent frameworks which researchers and practitioner use to deploy and produce RL models. I also attach a short roadmap for RL and using that I recently got into utilizing RL for the financial sector.
What is the Tensor Processing Unit (TPU)?
Here, I focused on the future of deep learning — Tensor Processing Units. The blog covers the initial introduction of TPUs, how it holds against GPUs and then the advantages in selective scenarios. The blog also provides a short practical implementation on colab using its TPU Functionality.
While I focus on vision problems of deep learning this experience helped me understand various domains of Machine Learning and Artificial Intelligence which I otherwise would have neglected. Writing blogs helps one increase the subject matter knowledge and develop written communication which is very important for any job sector
— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —
Feel free to send me connection request and message at Linkedin to collaborate on the above mentioned projects or a few tips on writing blogs. I also focus on CNNs for research so we could always have a chat on that!
Stay safe!
