Harikrishnam Raju’s Post

💼📊 Excited to share my latest diagram shedding light on the intricacies of neural network operations! Explore the mechanisms of forward and backward propagation, fundamental to the functioning of machine learning models. Let's delve into the world of AI and deepen our knowledge of advanced algorithms together. 🚀💡 #NeuralNetworks #ForwardPropagation #BackwardPropagation #AIInsights #DeepLearning

In March, Yann LeCun was hosted by NYAS for the Tata Series on AI & Society fireside chat. Dr. LeCun's groundbreaking work spans three decades, from his post doctoral work in Toronto, to current faculty role at NYU. He is the recipient of the Turing award. As both computer and mathematics scientist, he's led advances in AI ML, deep learning, computer vision, computational neuroscience. On avoiding AI monopolies: "open sourcing lessens the likelihood of a single company developing a monopoly over a particular technology. ...a single company simply does not have the ability to fine tune an AI system that will adequately serve the entire population of the world." #artificialintelligence #opensource #machinelearning https://lnkd.in/gwawn8Hh

"Embarking on an Exciting Journey: Sharing My Debut ANN Project on LinkedIn! 🌟.Unraveling the Types, Components, and Mathematics behind ANNs. 🧠✨ . Delighted to present my first venture into the world of Artificial Neural Networks. This project represents not only a milestone in my learning journey but also a testament to my passion for AI and its potential to revolutionize industries. Join me as I share the exhilaration of discovery, the challenges of experimentation, and the satisfaction of seeing ideas come to life. Let's connect, engage, and inspire one another as we navigate the fascinating landscape of ANN technology together! #ANN #AI #NeuralNetworks #MachineLearning #DeepLearning #DataScience"#dataanalytics #mathematics

Unlocking the Power of Complexity: CAS in AI Ever wondered how AI can adapt to the ever-changing world? The answer lies in complex adaptive systems (CAS). These are systems composed of interconnected agents that can learn and evolve, much like the natural world. CAS are crucial in AI because they: Mimic real-world complexity: Providing a more accurate representation of our dynamic world. Enable autonomous learning: Allowing AI to adapt to unforeseen challenges. Help predict complex behaviors: Understanding and anticipating the outcomes of complex systems. Inspire new AI architectures: Leading to innovative approaches like neural networks and genetic algorithms. Are you intrigued by the potential of CAS in AI? Let's discuss how we can harness this power to create more intelligent and adaptable AI systems. #AI #CAS #MachineLearning #ArtificialIntelligence #Complexity #Innovation

using only 19 control neurons, new intelligent system

Day 41 of #100DayChallenge, and on this vibrant Holi, let's celebrate a function as colorful as the festival itself - the ReLU (Rectified Linear Unit) activation function! The ReLU function, like a good filter in image processing, removes negativity while preserving the positive aspects of your data. In simpler terms, it allows only positive values to pass through, essentially converting negatives to zeros. Why is ReLU so popular? Here's a splash of its benefits: Speed & Efficiency:Compared to other activation functions, ReLU is computationally inexpensive, making it ideal for large neural networks. ⚡ Vanishing Gradient Problem:ReLU helps combat the vanishing gradient problem, a common hurdle in deep learning, by allowing gradients to flow freely through positive values. Biological Plausibility: Similar to how neurons fire in the brain, ReLU offers a more biologically realistic approach to artificial neural networks. So, this Holi, let's celebrate the power of ReLU to bring positivity and efficiency to our AI endeavors! #100DaysOfCode #ArtificialIntelligence #MachineLearning #DeepLearning

Excited to share my latest endeavor on LinkedIn! 🚀 I've embarked on a journey into the realms of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models. My goal? Building a cutting-edge Womens violence detection system. 🕵️♂️🔍 I'm thrilled to announce a significant milestone in my progress: my model can now detect specific actions like 'Slap 🖐 ' with impressive accuracy. Harnessing the power of these advanced AI technologies, I'm diving deep into the nuances of image and sequence analysis, exploring ways to enhance safety and security in our communities. Stay tuned for updates on this innovative project! #AI #DeepLearning #ViolenceDetection #ProgressUpdate

Interacting systems is a very interesting concept. While it may not always be possible to "reason" about interacting systems but using interacting systems we have been able to solve some of the greatest problems of the century. Take the below examples: 1. Neural Networks 2. Quantum Entanglement 3. Generative Adversarial Networks (the idea behind generative AI) ... and many more All of the above are interacting systems. In NN, neurons interact with each other in a way we cannot comprehend but they can approximate any function. In QE, quantum states of photons interact with each other to make computations exponentially faster. In GANs, generative network and adversarial networks interact with each other to improve outputs from generative networks. On a high level we know how the interaction happens but at a lower level they are still a mystery. For e.g. we still cannot reason about how many neurons or how many layers is sufficient to approximate a function. It's still "hit and miss". One of such interesting phenomena that exists in nature is the 3-body problem or the N-body problem. Although we can reason about 2 physical bodies or objects will interact but once we go to 3 bodies or higher, there is no closed form solution. My hunch is that in order to achieve AGI or solve NP-hard problems there could be interacting systems still not explored that will allow us to inch closer . #neuralnetworks #artificialintelligence #agi #nbodyproblem

Matbench Discovery: Ranking machine learning models based on their ability to find new, stable inorganic crystals quickly and efficiently. In materials science, the intersection of machine learning (ML) and crystal stability prediction has opened new frontiers for materials discovery and innovation. In this context, Janosh Riebesell and coauthors created "Matbench Discovery," an evaluation framework designed  together make it easy to rank ML energy models on a task designed to simulate a high-throughput discovery campaign for new stable inorganic crystals. "Matbench Discovery" ranks various ML methodologies, from random forests to graph neural networks and beyond, in their ability to predict crystal stability. Their work includes an evolving online leaderboard that showcases the trade-offs between different performance metrics. Up to today, ranked best-to-worst by their test set F1 score on thermodynamic stability prediction, the top 3 models in the ranking are: 1) GNoMe (https://lnkd.in/dJsvH23e) 2) MACE (https://lnkd.in/dN6HJQv6) 3) CHGNet (https://lnkd.in/dTDbhxvU) Link to MatBech Discovery site: https://lnkd.in/d9FVPM9t #Materials #MachineLearning #InorganicCrystals #HighThroughputScreening #AI #MLModels #MaterialsInformatics #CrystalDiscovery

Researchers at MIT have developed a groundbreaking approach that utilizes generative AI to tackle complex physics questions. By leveraging machine learning algorithms, they can generate novel solutions to intricate problems, revolutionizing the field. This method enables scientists to explore phenomena beyond conventional boundaries, opening new avenues for discovery and innovation. Through collaboration between AI and human expertise, the project marks a significant leap forward in scientific inquiry. The integration of AI into physics research promises to accelerate progress and uncover hidden insights in the universe's fundamental laws. This innovative approach underscores the transformative potential of AI in advancing our understanding of the natural world. #AI #ArtificialIntelligence #TechNews #technologynews #itconsultancy #ITservices #ukvisa #visasponsorship #Salient