Training an LLM is Like Growing a Fractal: The Power of Iterative Learning

Pooja Kashyap

James Gleick, one of my favorite science writers, wrote this famous quote in his book Chaos: Making a New Science:  “IN THE MIND’S EYE, a fractal is a way of seeing infinity”.’  Doesn’t it also reflect how LLMs work?  How? Just like fractals, these models take patterns, layer them and build something infinitely complex from simple rules. It’s like staring into a never-ending web of words, where meaning keeps unfolding the deeper we go.

Read More

Learning Together and Alone: How We Dynamically Adjust Our Strategies

Pooja Kashyap

Have you ever found yourself trying to solve a problem on your own, sticking with your gut and your own trial-and-error? But then there are times when you just can’t figure it out and end up watching what others are doing, hoping to pick up some clues? That back-and-forth between figuring things out solo and learning from others is what this study really dives into. It looks at how people balance learning on their own versus copying others depending on how well they’re doing.

Read More

Universe Is an Ultimate Computer: What if Space-Time is Pixelated?

Pooja Kashyap

I’m always on the lookout for some radical take when it comes to research papers, and this time, a research article by physicist Melvin Vopson totally caught my attention. Vopson suggests, what if gravity isn’t some deep, fundamental force of nature, but actually just the universe’s way of managing its information better? Sounds strange at first, but it’s got a logic to it.

Read More

Smarter Machines, Bigger Shifts: The New Era of Robotics

Pooja Kashyap

We’ve heard a lot about ChatGPT and generative AI, but there’s another side of AI that’s moving fast, which is, robots that can think and work on their own. These are smarter machines that can learn, make decisions, and adapt to their environment. And most of these machines are already making their presence felt in places like warehouses, factories, and even retail stores.

Read More

Supernova Remnant in Nearby Galaxy Studied via Radio

Pooja Kashyap

Astronomers just took a fresh look at a supernova remnant called MC SNR J0519–6902 tucked away in the Large Magellanic Cloud, a small galaxy orbiting the Milky Way. It’s been known since the early ’80s, but there’s still a lot we don’t fully understand about it, especially when it comes to its age and how it’s evolved over time.

Read More

DMS and DMDS Hints in K2-18 b’s Atmosphere from JWST MIRI Data

Pooja Kashyap

Recent research paper published by the American Astronomical Society – New Constraints on DMS and DMDS in the Atmosphere of K2-18 b from JWST MIRI – digs into the atmosphere of K2-18 b, which is the exoplanet people have been calling a “Hycean World,” meaning it might have a hydrogen-rich atmosphere and liquid water oceans. This study is based on data from JWST’s MIRI (Mid-Infrared Instrument), and it’s the first time we’ve gotten a mid-infrared transmission spectrum of a habitable-zone exoplanet like this.

Read More

Book Review: Zero to Birth by William A Harris

Pooja Kashyap

If you’re even a little bit curious about how a brain builds itself before we’re born, Zero to Birth: How The Human Brain Is Built is a book you’ll want to spend time with. It was first published in 2022 and is written by neuroscientist William A. Harris. What makes it so compelling is how clearly it lays out the stages of brain development without dumbing things down.

Read More

Interview: Dr. Nicholas Jacobson, Computational Psychiatrist at Dartmouth College, US

Pooja Kashyap

When I first came across Therabot, Dartmouth’s AI-powered therapy chatbot, I was immediately intrigued. The idea that technology could play a meaningful role in mental health treatment fascinated me, so I reached out to Dr. Nicholas Jacobson, a leading expert in this space, for an interview. To my delight, he agreed.

Read More

Artemis-Enabled Stellar Imager: A Lunar Telescope Array to Unveil the Secrets of Stars and Black Holes

Pooja Kashyap

Imagine building a huge group of telescopes on the Moon, where there’s no air or weather to interfere, allowing us to see further into space than we ever could before. That’s the idea behind the Artemis-enabled Stellar Imager (AeSI), a proposed lunar interferometer designed to reveal what’s happening on the surfaces of stars and deep inside active galactic nuclei (AGN). A team led by Dr. Kenneth Carpenter at NASA’s Goddard Space Flight Center recently completed a nine-month feasibility study, and their findings suggest that this ambitious idea is not just…

Read More

AntGrip: Enhancing Gripper Performance with Ant-Inspired Hairs

Pooja Kashyap

Whenever I find myself really fascinated by biomimicry, I often discover new things inspired by it. This time, it’s a robotic hand that takes ideas from how ants work. Imagine you’re picking up a slippery soda can with a simple two-finger robot gripper. No suction cups, no fancy sensors, just friction. Sounds tricky, right? That’s exactly the kind of challenge researchers tackled by taking inspiration from “ants”.

Read More

Infomorphic Neurons: Self-Learning AI Inspired by Biology

Pooja Kashyap

Imagine if artificial neurons could learn like the ones in our brains, adapting, making decisions on their own, and figuring things out without needing a central command. That’s exactly what scientists are working on with a new kind of artificial neuron called infomorphic neurons. Unlike traditional artificial neural networks, these neurons don’t need an external system to guide their learning. Instead, they learn in a self-organized way, just like biological neurons do.

Read More