We reimagined Transformer architectures inspired by nature's hidden structures

> This paper proposes a novel approach grounded in a fourfold pattern observed at multiple levels of granularity in nature, referred to as ‘ Kn′, ‘Po’, ‘Pr’, and ‘H’, which correlates with concepts of knowledge, presence, power, and harmony.

Triggers my crackpot radar. One of the authors has a bunch of other solo papers (ostensibly) about quantum mechanics that look similarly nuts, like this one:

https://ieeexplore.ieee.org/author/37086022058

> Hence, looking at Euler's Identity which comprises of numbers between the range ‘0’ to ‘4’, from the point of view of light, it becomes possible to understand quantum-level dynamics as an outcome of a multi-layered, symmetry-based model of light.

What the drugs.

It's a bit weird that such a paper (standard ML topic, would fit to NeurIPS, ICLR, ICML or so) is published in such a conference (IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)), which seems not really the best fit. This indicates that the quality bar would not be reached for the top venues.

If you want to read some recent interesting ML papers (on Transformers, or whatever you care), check e.g. recent ICLR 2025 papers (https://openreview.net/group?id=ICLR.cc/2025/Conference#tab-...), or other recent venues.

> This paper proposes a novel approach grounded in a fourfold pattern observed at multiple levels of granularity in nature, referred to as ‘ Kn′, ‘Po’, ‘Pr’, and ‘H’, which correlates with concepts of knowledge, presence, power, and harmony.

Sounds a lot like semi-profound bullshit to me..

Given that the IEEE stands for the Institute of Electrical and Electronics Engineers, you should be forgiven for thinking that this would be about transformers and the electric grid.

paywalled. got a link?