Placebo Effect Grows in U.S., Thwarting Development of Painkillers
https://www.scientificamerican.com/article/placebo-effect-gr...
The most interesting thing is that "placebo responses are rising only in the United States."
Ask HN: Which recent research paper blew your mind?
81–90 of 174 posts
Re: Ask HN: Which recent research paper blew your mind?
#82Integral Neural Networks (CVPR 2023 Award Candidate), a nifty way of building resizable networks. My understanding of this work: A forward pass for a (fully-connected) layer of a neural network is just a dot product of the layer input with the layer weights, followed by some activation function. Both the input and the weights are vectors of the same, fixed size. Let's imagine that the discrete values that form these…
Going just by your description this sounds like they are doing operator learning. It's actually a very old idea. The proof that started operator learning is from 1988 I believe. Mathematicians have been playing around with the idea since 2016 at least.
Indeed, this seems closely related, thanks for the pointer!
Unfortunately I'm not deep enough into the topic to understand what their contribution to the theory part of it is. (they have some Supplementary Material in [INN Supp]). In the discussion of the Integral Neural Networks (INN) paper, there's this paragraph about an operator learning publication:
"In [24] the authors proposed deep neural networks with layers defined as functional operators. Such networks are designed for learning PDE solution operators, and its layers are continuously parameterized by MLPs only along the kernel dimensions. A re-discretization was investigated in terms of training on smaller data resolution and testing on higher input resolution. However, the proposed framework in [24] does not include continuous connections between filters and channels dimensions."
Also the weight permutation to perform the resampling on pretrained networks in INNs seems to be novel? And I guess it doesn't hurt that they're bringing new eyeballs to the topic, by providing examples of common networks and a PyTorch implementation.
[INN Supp]: https://openaccess.thecvf.com/content/CVPR2023/supplemental/...
[24]: Zongyi Li Nikola Kovachki. Neural operator: Graph kernel network for partial differential equations. arXiv preprint arXiv:2003.03485, 2020, https://arxiv.org/abs/2003.03485
Re: Ask HN: Which recent research paper blew your mind?
#83Someone managed to GPU-accelerate program synthesis, a form of symbolic ML. First time for ML that is not deep learning: https://dl.acm.org/doi/10.1145/3591274 Deep learning took off precisely when the ImageNet paper dropped around 2010. Before nobody believed that backprop can be GPU-accelerated.
That's cool and all, but the one thing that really made it possible to train deep neural nets was the use of backpropagation, and its polynomial time complexity.
By contrast, there are no known polynomial time algorithms for program synthesis and the standard approach is to search some large combinatorial space [1]. That's the case for all the classical approaches: SMT, SAT, planning and scheduling, etc. At the same time there's very powerful heuristics for all the other classical problems that can solve many problem instances efficiently.
____________
[1] The one exception to this is Inductive Logic Programming, i.e. the inductive synthesis of logic programs, for which we do know a polynomial time algorithm (but that is my work so I'm not pimping it here).
Re: Ask HN: Which recent research paper blew your mind?
#84Not all of these are research papers. But all are fairly recent. Gene linked to long COVID found in analysis of thousands of patients https://www.nature.com/articles/d41586-023-02269-2 Surfactants safely take down mosquitoes without using insecticides https://newatlas.com/science/surfactants-safely-take-down-mo... This is what our Milky Way galaxy looks like when viewed with neutrinos https://arstechnica.com/science/…
>Surfactants safely take down mosquitoes without using insecticides https://newatlas.com/science/surfactants-safely-take-down-mo...
Curious - what would happen to a spider or an ant that ate a couple of mosquitoes with that spice on top? Will they also suffocate?
Re: Ask HN: Which recent research paper blew your mind?
#85"Co-cultivation enhanced microbial protein production based on autotrophic nitrogen-fixing hydrogen-oxidizing bacteria" https://www.sciencedirect.com/science/article/abs/pii/S13858... Certain bacteria can directly assimilate a mixture of hydrogen, carbon dioxide, and nitrogen to produce protein. You could consider it an alternative to bacterial nitrogen fixation in root nodules with much higher productivity. Or you c…
We hear that eating vegetables is more efficient in ecological footprint than eating meat, since it cuts out the middle man. Is it yet more efficient to cut out the plants and get dietary protein from the bacteria that feed them?
Possibly, yes. This is the dream of so called "single cell protein" production. One type of SCP has been sold for years under the brand name Quorn (derived from a fungus rather than bacteria).
Bacterial protein may trigger allergic reactions in people and bacterial biomass is purine-rich which can also be a problem for people prone to gout. It's possible that cell engineering, directed evolutionary selection, or additional post-growth processing can minimize these problems.
I personally think that the more likely path is using fast-growing bacteria as feed for animal agriculture or aquaculture. Solar panels are so efficient at sunlight conversion compared to plants that you could farm salmon protein starting from bacterial pellets grown on solar derived hydrogen with per-hectare productivity comparable to conventionally farming soy beans. But the solar farm can go on saline, dry, contaminated, or otherwise agriculturally useless land. And salmon has slightly greater nutritional value than soy protein plus significantly greater market value.
Re: Ask HN: Which recent research paper blew your mind?
#86"Overview of SHARD: A System for Highly Available Replicated Data" it's the first paper to introduce the concept of database sharding. It was published in 1988 by the Computer Corporation of America. It is referenced hundreds of times in many classic papers. But, here's the thing. It doesn't exist . Everyone cites Sarin, DeWitt & Rosenb[e|u]rg's paper but none have ever seen it. I've emailed dozens of academics, libr…
Here's something that seems related. Maybe one of these authors would have a copy of the other paper? Not sure if they would be among the set of folks you've already tried or not... https://apps.dtic.mil/sti/pdfs/ADA171427.pdf
Interesting, thanks. Wonder if it was a coincidence that UO adopted that terminology, or if somebody there knew about this.
Re: Ask HN: Which recent research paper blew your mind?
#87"Liquid solution centrifugation for safe, scalable, and efficient isotope separation" https://www.science.org/doi/10.1126/sciadv.adg8993 The authors show that a biological type laboratory ultracentrifuge can efficiently function as a near-universal isotope separator. Any element that can be dissolved as a salt in water -- the entire periodic table, excepting the noble gases -- can be enriched according to its relativ…
Surely something about the technique would make it impractical for enriching uranium though... you're not going to be able to produce kilograms of 20% U235 with this are you?
Re: Ask HN: Which recent research paper blew your mind?
#88Earlier quoted context omitted.
1988? As far as anyone can tell, the use of the term "shard" in the context of database replication originated with Ultima Online, which was released in 1997, and which used the term in connection with its underlying mythos (the idea of representing world instances as shards of Mondain's shattered gem). So a documented reference to sharding that's earlier than that would be interesting to see. (Disagree? Instead of d…
You can find reports from before 1988 mentioning SHARD being in development, like the one from June 1986 linked in this sister comment: https://news.ycombinator.com/item?id=36849634
Some more, from 1989[1][2][3]. Which again, reference the missing "SHARD" paper, but contain enough detail to make it clear that the idea of SHARD existed, regardless of the status of that particular document.
[1]: https://apps.dtic.mil/sti/tr/pdf/ADA214478.pdf
Re: Ask HN: Which recent research paper blew your mind?
#89"Overview of SHARD: A System for Highly Available Replicated Data" it's the first paper to introduce the concept of database sharding. It was published in 1988 by the Computer Corporation of America. It is referenced hundreds of times in many classic papers. But, here's the thing. It doesn't exist . Everyone cites Sarin, DeWitt & Rosenb[e|u]rg's paper but none have ever seen it. I've emailed dozens of academics, libr…
Huh. OK, here's something that might be interesting. I found another paper[1] that cites SHARD, but the citation is slightly different. Instead of being a CCA memo, it shows it as a Xerox memo:
Sunil Sarin, Mark DeWitt, and Ronni Rosenberg, "Overview of SHARD: A System for Highly Available Replicated Data," Technical Report 162, Xerox Advanced Information Technology (May 1988).
EDIT:
OK, I think I get this now. I had read the Wikipedia blurb about CCA being acquired by Rocket earlier, but only just now did I keep reading further down to find this bit:
in 1984, CCA was purchased by Crowntek, a Toronto-based company.[8] Crowntek sold Computer Corporation of America's Advanced Information Technology division to Xerox Corporation in 1988.[9] The balance of CCA was acquired by Rocket Software, a Boston-based developer of enterprise infrastructure products,[2] in April 2010.
So it seems like the portion of CCA that would be of interest here, is probably the bit that sent to Xerox. Maybe somebody at Xerox could help turn up the missing document?
I doubt it will help, but I took a stab at pinging them on Twitter X.
Re: Ask HN: Which recent research paper blew your mind?
#90Someone managed to GPU-accelerate program synthesis, a form of symbolic ML. First time for ML that is not deep learning: https://dl.acm.org/doi/10.1145/3591274 Deep learning took off precisely when the ImageNet paper dropped around 2010. Before nobody believed that backprop can be GPU-accelerated.
That's cool and all, but the one thing that really made it possible to train deep neural nets was the use of backpropagation, and its polynomial time complexity. By contrast, there are no known polynomial time algorithms for program synthesis and the standard approach is to search some large combinatorial space [1]. That's the case for all the classical approaches: SMT, SAT, planning and scheduling, etc. At the same…
I'm working on an inductive logic programming algorithm, and from my understanding the search space for logic programs is just as vast as other types of program synthesis. Do you have any information you can share about the polynomial time algorithm?