Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1038/scientificamerican1018-4 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Opening the AI Black Box: Distilling Machine-Learned Algorithms into Code.
Entropy (Basel)
December 2024
Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
Can we turn AI black boxes into code? Although this mission sounds extremely challenging, we show that it is not entirely impossible by presenting a proof-of-concept method, MIPS, that can synthesize programs based on the automated mechanistic interpretability of neural networks trained to perform the desired task, auto-distilling the learned algorithm into Python code. We test MIPS on a benchmark of 62 algorithmic tasks that can be learned by an RNN and find it highly complementary to GPT-4: MIPS solves 32 of them, including 13 that are not solved by GPT-4 (which also solves 30). MIPS uses an integer autoencoder to convert the RNN into a finite state machine, then applies Boolean or integer symbolic regression to capture the learned algorithm.
View Article and Find Full Text PDFOrdering Effect of Charge-Charge Repulsion in Doped Antiferromagnetic Lattices: A Coupled Cluster Study.
J Phys Chem A
November 2024
Laboratoire de Chimie et Physique Quantiques, Université Toulouse III─Paul Sabatier, Toulouse 31062, France.
In quantum chemistry, single-reference Coupled Cluster theory, and its refinements introduced by Bartlett, has become a "gold-standard" predictive method for taking into account electronic correlations in molecules. In this article, we introduce a new formalism based on a Coupled Cluster expansion of the wave function that is suited to describe model periodic systems and apply this methodology to the case of hole-doped antiferromagnetic two-dimensional (2D)-square spin-lattices as a proof of concept. More precisely, we focus our study on 1/5 and 1/7 doping ratios and discuss the possible ordering effect due to large hole-hole repulsion.
View Article and Find Full Text PDFTracking the progeny of bacterial persisters using a CRISPR-based genomic recorder.
Proc Natl Acad Sci U S A
October 2024
Department of Microbiology, Harvard Medical School, Boston, MA 02115.
The rise of antimicrobial failure is a global emergency, and causes beyond typical genetic resistance must be determined. One probable factor is the existence of subpopulations of transiently growth-arrested bacteria, persisters, that endure antibiotic treatment despite genetic susceptibility to the drug. The presence of persisters in infected hosts has been successfully established, notably through the development of fluorescent reporters.
View Article and Find Full Text PDFEmergency Medical Access Control System Based on Public Blockchain.
J Med Syst
September 2024
Faculty of Engineering, Information and Systems, University of Tsukuba, Ibaraki, Japan.
IT has made significant progress in various fields over the past few years, with many industries transitioning from paper-based to electronic media. However, sharing electronic medical records remains a long-term challenge, particularly when patients are in emergency situations, making it difficult to access and control their medical information. Previous studies have proposed permissioned blockchains with limited participants or mechanisms that allow emergency medical information sharing to pre-designated participants.
View Article and Find Full Text PDFA circadian behavioral analysis suite for real-time classification of daily rhythms in complex behaviors.
bioRxiv
August 2024
Department of Biology, Texas A&M University, College Station, TX.
Article Synopsis
- Measuring animal behavior has been challenging over long timescales, especially for complex behaviors that require manual scoring, but advances in machine learning offer new solutions.
- A new pipeline has been developed to continuously record and analyze animal behavior using a self-supervised computer vision model, achieving high accuracy in behavior classification comparable to expert human labelers.
- This system was tested on 97 mice over two weeks, revealing significant sex- and estrogen-dependent differences in circadian rhythms among various behaviors, and is presented as an open-source software tool called the "circadian behavioral analysis suite" (CBAS) for researchers.
Want AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!