iiinova — where your digital self keeps evolving

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 25/09/2026

000

So you're saying that there's no need because having their name on the paper implies that already?

120

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 27/08/2026

Thanks to Dave Clader for the many fruitful discussions and to Maria Schuld for her email exchange (both back in 2023!) On a side note, I wonder given the current state of AI if such notes will no longer be necessary in the near future...

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 27/08/2026

Happy to finally get this comment, mostly written in 2023, out on arXiv! If you care about quantum kernels, then I would recommend at least a quick read through section 1 :) arxiv.org/abs/2608.24923 As usual any feedback is welcome!

arxiv.org

Comment on 'Supervised quantum machine learning models are kernel methods'

We identify a few small errors in the proof of Theorem 1 in M. Schuld, 'Supervised quantum machine learning models are kernel methods' (arXiv:2101.11020v2), Appendix A. These do not affect the validit...

120

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 27/08/2026

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 01/08/2026

Agreed! But then the repo owner says > ...represents, conservatively, one to six months of focused work. At a postdoc's fully-loaded cost that is somewhere between $15,000 and $80,000, and the leading constraint isn't even the money

120

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 31/07/2026

Same! Also added my one cent 😅 great to see thay many people are thinking about this openly!

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

five :) it's written in the first comment (i had to make a thread due to the post character limit)

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

A big shout out to my co-authors at QuantumBasel and Moonlight AI - Julien Baglio, Frederik Flöther. Christian Ruiz, Stefan Habringer and Nicole H. Romano for a really fun and fruitful collaboration!

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

standardizing highly efficient algorithms for widespread, real-world precision medicine is an ongoing pursuit.

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...to do there! -The quantum computing integration was conducted as a "proof of concept," the scaling, refinement, and full integration of these quantum algorithms into daily clinical diagnostic tools remains an open challenge. -The bigger goal is toward single-cell diagnostics, and...

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...it successfully improves the AI's sensitivity while dramatically reducing the computing power and complexity needed to train it. -We also performed a successful proof of concept demonstrating how to enrich this classical computing framework with quantum algorithms, but still more work...

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...(a >4% drop in AUC). -The main method was attention-based multiple instance learning (MIL) -We introduced a novel algorithm that combines extreme learning machines with attention-based MIL, and also introduced a new method using higher-dimensional feature spaces. This is important because...

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...a factor of 5, making the AI lighter to train, while maintaining an average AUC within just 1.5% of the heavier, traditional deep learning model. - We also found that nonlinearities are vital; removing them led to a sharp decrease in both model stability and overall performance...

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...makeup. -By analyzing images of different blood cells in more complex ways (higher-dimensional feature spaces), we achieved an increase in the model's robustness, improving the AUC by over 10%. -The implementation of extreme learning machines reduced the number of trained parameters by...

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...challenge in computational pathology. -Solving this problem is important because highly accurate and efficient single-cell diagnostics are foundational building blocks for precision medicine, a field dedicated to tailoring treatments to individual patients based on their unique cellular...

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

...five and boosted its robustness by over 10%. We also implemented a proof-of-concept test integrating quantum algorithms, but still more work to do there! More details: -Analyzing whole-slide images to detect circulating rare cells, such as erythroblasts, in peripheral blood is a major...

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/07/2026

New paper out in ML: Health (IOP Publishing) lnkd.in/dMRvkHxg TL;DR: We developed an AI model for detecting rare blood cells in medical images by combining "extreme learning machines" with higher-dimensional feature spaces. This approach reduced the training complexity by a factor of...

lnkd.in

Checking your browser - reCAPTCHA

200

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 23/06/2026

Nice! What's material for the one titled "Google" (different to the wormhole one I presume?)

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 01/06/2026

Did they guess it or could it have been using a sneakier way like using a keyboard logger or looking over your shoulder?

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 14/05/2026

For those interested in the alternate measures that are mentioned near the bottom of the petition: uonucu.org/wp-content/uploads/2025/07/UCU-Phase-One-Counterproposal.pdf

uonucu.org

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/04/2026

Haha fair enough 😅 genuine question: you're not so enthusiastic about some of the recent work on spectral methods (e.g. arxiv.org/abs/2603.24654), QELM/QRC (e.g. arxiv.org/abs/2603.13005) or hybrid QML (e.g. arxiv.org/abs/2504.16275) being maybe useful on NISQ (none of these works are mine)?

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 30/04/2026

Test analogy on NISQ vs Fault-Tolerant (FT). NISQ is easy to get started on but unclear ultimate potential - like path that is initially easy but the top is clouded. FT has a clear(er) path, but tougher to start and still some known and unknown challenges before it's useful. Feedback/thoughts?

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 10/04/2026

This one is very interesting! Many discussions already happening on scirate!

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 01/04/2026

You're quoting a comment on scirate by "qanon", whereas the paper states it will take several months.

Tom Rivlin @tomrivlin.com · 25/03/2026

Reposted by Rajiv Krishnakumar

New SciShow I wrote! What's the strongest laser in the world? Well, it depends what you mean by strongest... Find out more in this super fun video! Also featuring possibly my favourite introduction I've ever written for a video www.youtube.com/watch?v=gqPl...

youtube.com

The 3 Strongest Lasers in the World (For Now)

YouTube video by SciShow

121

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 11/03/2026

Interesting that they don't discuss quantum annealing (I have no strong opinions on whether it should or shouldn't be included)

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 08/03/2026

Thank you for sharing! Is there per chance also a recording of the talk available online?

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 08/03/2026

A really lively discussion with the quantum mathematician Devika Sharma! Ranging from error correction topics to taking care of oneself during her free time, Priya and I really enjoyed a bit of insight into what the life of a quantum mathematician is like 😊 youtu.be/maXkByXOOuU

youtu.be

A Day in the Life of a Quantum Mathematician with Devika Sharma | Ep. 9

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/02/2026

A big shout out to the first author Sparsh Mittal along with Mintu Kumar and Yash Chand for their valuable contributions, and of course Prof. Neel Kanth Kundu for orchestrating the work

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/02/2026

We tested this using indoor localization data (finding a person's location inside a building using signals from WiFi, Bluetooth, and Zigbee)

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/02/2026

The (small but clean) novel contribution that our paper shows is that you can use a very naive method of plucking a few integer Fourier frequencies around zero (e.g. [-1, 0, 1]) as a heuristic and it worked for the cases we tried!

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/02/2026

In more detail - we used the already-known method of taking the partial Fourier series of the quantum layer and keeping only the leading terms. We saw that this did not degrade the performance of the model.

100

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/02/2026

Paper on taking the classical surrogate of QML models! ieeexplore.ieee.org/abstract/doc... Taking a classical surrogate means we took a hybrid QML model and translated the quantum layer to a "classical surrogate" so that the whole model can be run efficiently on a classical computer.

ieeexplore.ieee.org

Experimental Validation of Dequantization of Hybrid Quantum Machine Learning Models Using Classical Surrogates

Dequantization of quantum machine learning models is the methodology of creating an equivalent classical model using a partial Fourier series over the classical input variables such that the classical...

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 12/02/2026

This feels similar to when companies started putting wifi to many household appliances... adding risk with no practical benefits

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 21/01/2026

Could the new restrictions be to try and combat the AI slop?

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 24/12/2025

And that's another episode in the bag! This one with Kishor was really nice to hear about his days and his philosophy on how he works on quantum science, and what it takes (including making time for the really important things like family) 😁 www.youtube.com/watch?v=zqm-...

youtube.com

A Day in the Life of a Quantum Information Theorist with Kishor Bharti | Ep. 8

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 08/12/2025

It was really fun working with Julian Jang-Jaccard and everyone at the Cyber-Defense Campus🇨🇭and a shout out to my boss Frederik Flöther for his comments on shaping this chapter

000

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 08/12/2025

The whole book is very understandable to anyone and is 𝗳𝘂𝗹𝗹𝘆 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲! So if you or anyone you know is interested to learn more about how quantum computing can impact them personally or professionally, I humbly recommend this chapter 🙂

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 08/12/2025

Very excited that the book 𝗤𝘂𝗮𝗻𝘁𝘂𝗺 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀: 𝗧𝗿𝗲𝗻𝗱𝘀 𝗮𝗻𝗱 𝗜𝗺𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗳𝗼𝗿 𝗖𝘆𝗯𝗲𝗿 𝗗𝗲𝗳𝗲𝗻𝘀𝗲 is finally out, which includes my chapter on 𝗤𝘂𝗮𝗻𝘁𝘂𝗺 𝗧𝗵𝗿𝗲𝗮𝘁𝘀: link.springer.com/chapter/10.1...

link.springer.com

Quantum Threats

With the recent advancements in quantum computing, there is a threat that today’s widely used digital encryption schemes will become vulnerable to cryptography-breaking quantum algorithms. In pa...

130

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 19/08/2025

Really excited to share this chat with Awanish Pandey, an optics and photonics professor at IIT Delhi. Very insightful and fun conversations with a great sense of humour all along :) And of course a shout out to my co-host Priya! Stay tuned for more episodes! youtu.be/vwoY9OELZco?...

youtu.be

A Day in the Life of an Optics and Photonics Professor with Awanish Pandey

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 05/08/2025

Does this mean the coin flipping guys get to publish a paper too? (Sidenote: I am unironically a big fan of trickshot videos. The joy in their faces when they finally get it is fantastic!)

Rajiv Krishnakumar @rajivkrishnakumar.bsky.social · 17/07/2025

I think a way to fight blog posts is to preach to non-experts to 1. Find the link to the scientific paper in the blog post 2. Upload just the paper (not blog post or any other media) into NotebookLM 3. Ask NotebookLM your questions It's not perfect but (so far) it's pretty decent!