Nobel Prizes 2024

[lpetrich]

All Nobel Prizes 2024 - NobelPrize.org - I am sadly slow on this.

Press release: The Nobel Prize in Physics 2024 - NobelPrize.org
The two winners are John J. Hopfield and Geoffrey Hinton, “for foundational discoveries and inventions that enable machine learning with artificial neural networks”.

This year’s two Nobel Laureates in Physics have used tools from physics to develop methods that are the foundation of today’s powerful machine learning. John Hopfield created an associative memory that can store and reconstruct images and other types of patterns in data. Geoffrey Hinton invented a method that can autonomously find properties in data, and so perform tasks such as identifying specific elements in pictures.

Seems to me more like algorithm design, though they used some theoretical-physics models in their work.

For JH's work, it was using "spin glass theory", simplified models of magnetic materials that involve the interaction of each atom's magnetic field with its neighbors' magnetic fields. Simplified like the  Ising model. where each atom has field value +1 or -1 and each neighboring pair's interaction energy is proportional to the product of field strengths.

GH's work seems to involve "simulated annealing", much like hill climbing, but where one rejects wrong-way moves only some of the time. This enables the system to wander away from local minima until it finds a strong one.

 Neural network (machine learning) - inspired by how biological neurons work. They use cascaded and thresholded linear classifiers, thus getting around the limitations of linearity.

Limitations of linearity? There are many problems that cannot be solved with linear classifiers, even thresholded ones, including some very simple ones, like the exclusive-or problem:

0 0: 0
0 1: 1
1 0: 1
1 1: 0

Plain or is easy to solve with a linear classifier, however. Add the inputs then threshold that sum: x = min(x1+x2,1)[/quote]

 

[lpetrich]

Press release: The Nobel Prize in Chemistry 2024 - NobelPrize.org

1/2 to David Baker “for computational protein design” and 1/4 each to Demis Hassabis and ohn Jumper “for protein structure prediction”

Proteins? These molecules are strings of "amino acids", each one with an "amino group", -NH2, and a "carboxylic acid group", -COOH. The AA's are connected by joining their amino and carboxyl groups. Organisms use a set of 20 protein-forming amino acids with a variety of chemical properties, and not surprisingly, this can produce a great variety of proteins.

Proteins form enzymes, biological catalysts, and also membrane pumps and structural material. Immune systems recognize their targets by responding to some of their proteins sticking to those stargets.

Not surprisingly, to work correctly, proteins' amino-acid strands must have the proper 3D arrangement for those proteins to function. But predicting protein structures from their AA sequences is a very difficult task, and DH and JJ presented an AI model called AlphaFold2 that was capable of predicting nearly every known structure. AlphaFold Protein Structure Database

How can one observe the structure of something much smaller than a wavelength of visible light? With something with a similar-sized wavelength: X-rays. One does X-ray crystallography, making a crystal of what one's researching, and then using that crystal as a diffraction grating for X-rays. From where one sees the X-rays emerge, one can work backward to find the structure of that crystal, and that includes where all the AA's are in a protein.

Once one can predict a protein's structure, one can use that ability to design proteins with desired structures, and that is what DB got his part of the prize for.

 

[lpetrich]

Press release: The Nobel Prize in Physiology or Medicine 2024 - NobelPrize.org
To Victor Ambros and Gary Ruvkun "for the discovery of microRNA and its role in post-transcriptional gene regulation"

Most known gene regulation is of gene expression, copying of genome DNA to messenger RNA. But what VA and GR discovered was another kind: bits of RNA that attach to messenger RNA. The first microRNA, lin-4 was discovered in the soil and lab nematode Caenorhabditis elegans But was it a quirk of these 1-mm-long worms? The second one, let-7 was one with a highly-conserved gene spread all over the animal kingdom.

Since then, hundreds of these regulatory RNA's have been discovered, comparable to the number of known "transcription factors", which the DNA-to-RNA step.

 

[lpetrich]

The Prize in Economic Sciences 2024 - Press release - NobelPrize.org - not a true Nobel Prize, but a Nobel Memorial Prize.

Daron Acemoglu, Simon Johnson, and James A. Robinson “for studies of how institutions are formed and affect prosperity”

Societies with a poor rule of law and institutions that exploit the population do not generate growth or change for the better. ...

When Europeans colonised large parts of the globe, the institutions in those societies changed. This was sometimes dramatic, but did not occur in the same way everywhere. In some places the aim was to exploit the indigenous population and extract resources for the colonisers’ benefit. In others, the colonisers formed inclusive political and economic systems for the long-term benefit of European migrants.

...
Some countries become trapped in a situation with extractive institutions and low economic growth. The introduction of inclusive institutions would create long-term benefits for everyone, but extractive institutions provide short-term gains for the people in power. As long as the political system guarantees they will remain in control, no one will trust their promises of future economic reforms. According to the laureates, this is why no improvement occurs.