Recent Additions: New and Updated Entries in the Past 60 Days
Displaying entries 16 through 30 of 80 found.

Pedia Sapiens: A Planetary Progeny Comes to Her/His Own Actual Factual Knowledge

Alser, Mohammed, et al. Going from Molecules to Genomic Variations to Scientific Discovery. arXiv:2205.07957. We cite this entry by an eight person ETH Zurich team to record a dedicated project to access the latest deep learning techniques so as to achieve a realm of Iintelligent Algorithms and Architectures (hardware) for next generation sequencing needs.

A great need now exists to intelligently read, analyze, and interpret our genomes not more quickly, but accurately and efficiently enough to scale to population levels. Here we describe much improved genome studies by way of novel AI algorithms and architectures. Algorithms can access genomic structures as well as the underlying hardware. We move onto future challenges, benefits, and research directions opened by new sequencing technologies and specialized hardware chips. (Excerpt)

Chantada, Augusto, et al. Cosmological Informed Neural Networks to Solve the Background Dynamics of the Universe. arXiv:2205.02945. We cite this entry by five astro-analysts from Argentina and Harvard as an example of how 2020s AI (EI) techniques can achieve a epic advance (quantum leap) in analytic prowess as our collective Earthuman proceeds apace with this apparent task of ecosmic self-description. See also Stellar Mass and Radius Estimation using Artificial Intelligence by Andy Moya and R. Lopez-Sastre at 2203.06027, and What a neural network model learns about Cosmic Structure Formation by Drew Jamieson, et al at (2206.04573) for more usages.

The field of machine learning has drawn increasing interest due to its ability to solve many different problems. In this work, we train artificial neural networks to represent differential equations that govern the background dynamics of the Universe. We chose four models to study: ΛCDM, parametric dark energy, quintessence and the Hu-Sawicki f(R) model. We performed statistical analyses to estimate each model's parameters by observational data. We found that the error of the solutions was ∼1% in the region of the parameter space. (Excerpt)

Chen, Boyuan, et al. Discovering State Variables Hidden In Experimental Data. arXiv:2112.10755. This entry by Columbia University computer scientists led by Hod Lipson offers a good survey of how this computational endeavor began and goes forth today. It opens by noting historic studies of physical laws and motions as a search for elusive values. From 2021, it is advised that as not before bovel AI methods can achieve deeper analyses so as to discern their presence in dynamic systems such as reaction-diffusion. See also Distilling Free-Form Natural Laws from Experimental Data by Michael Schmidt and Hod Lipson in Science (324/5923, 2009, second Abstract).

All Physical laws are based on relationships between state variables which give a description of the relevant system dynamics. However, the process of identifying the hidden state variables has so far resisted AI techniques. We propose a new principle to find how many state variables an observed system is likely to have, and what these variables might be. Without any prior knowledge of the underlying physics, our algorithm discovers the intrinsic dimension of the observed dynamics and identifies sets of state variables. We suggest that this approach could help catalyze the understanding, prediction and control of increasingly complex systems. (Excerpt)

For centuries, scientists have attempted to identify and document analytical laws that underlie physical phenomena. Despite much computing power, the process of finding natural laws and their equations has resisted automation. We need to define an algorithm which can insightfully correlate observed data sets. Without prior knowledge about physics, kinematics, or geometry, our algorithm discovered Hamiltonians, Lagrangians, and momentum conservation.. (2009 Abstract)

Cranmer, Miles, et al. Discovering Symbolic Models from Deep Learning with Inductive Biases. arXiv;2006.11287. Seven Princeton U., Deep Mind, London, NYU, and Flatiron Institute, NYC computer specialists articulate yet another effective machine procedure as our learning (and hopefully thinking) planet begins to spiral up to a prodigious Earthropic sapiens phase.

We develop a general approach to distill symbolic representations of a learned deep model by introducing strong inductive biases. We focus on Graph Neural Networks (GNNs) that encourage sparse latent representations an apply symbolic regression to learned model components to extract physical relations. We go on to study a cosmology sample of detailed dark matter and are discover a analytic formula that can predict the concentration of dark matter from the mass distribution of nearby cosmic structures. Our approach offers new ways to interpret neural networks and revealing physical principles from their representations. (Abstract)

Higgins, Irina, et al. Symmetry-Based Representations for Artificial and Biological General Intelligence. arXiv:2203.09250. DeepMind, London really intelligent persons IH, Sebastien Racaniere and Danilo Rezende scope out ways that an intersect of computational frontiers with neuroscience studies can benefit each field going forward. Once again, an Earthificial realm becomes more brain-like as human beings may first program so that the algorithms can process their appointed tasks (if all goes to plan) and come up with vital contributions on their own.

Biological intelligence is remarkable in its ability to produce complex behaviour in diverse situations. An ability to learn sensory representations is a vital need, however there is little agreement as to what a good representation should look like. In this review we contend argue that symmetry transformations are a main principle. The idea these phenomena affect some aspects of a system but not others, has become central in modern physics. Recently, symmetries have gained prominence in machine learning (ML) by way of more data efficient and generic algorithms that mimic complex behaviors. Taken together, these symmetrical effects suggest a natural framework that determines the structure of the universe and consequently shapes both biological and artificial intelligences. (Abstract excerpt)

Kitano, Hiroaki. Nobel Turing Challenge: Creating the Engine for Scientific Discovery. NPJ Systems Biology. 7/29, 2021. A leading Japanese executive scientist who directs its Systems Biology Institute outlines a comprehensive, insightful project as it becomes more evident that AI computational algorithmic capacities, if properly informed and trained, can proceed to run programs, process data, iterate, and optimize research studies on their own. As since this frontier now involves many worldwise collaborations, as the spiral turns maybe a new collective group “Global Prize” in recognition would be appropriate. And this time it should include the missing life and mind sciences. See also herein entries by Charlie Wood as this Earthuman acumen gains momentum.

Scientific discovery has long been one of the central driving forces in our civilization. It uncovered the principles of the world we live in, and enabled us to invent new technologies reshaping our society, cure diseases, explore unknown new frontiers, and hopefully lead us to build a sustainable society. In this regard, we propose an overall “science of science” to guide and boost going forward. A prime facility into these 2020s thus need be a viable integration of artificial intelligence (AI) system. We are aware that the contributions of “AI Scientists” may not resemble human science but deep hybrid-AI methods could take us our cognitive limitations and sociological constraints. (Excerpt)

Krenn, Mario, et al. On Scientific Understanding with Artificial Intelligence. arXiv:2204.01467. Twelve scholars in Germany, Canada, the USA, and China including Alan Aspuru-Guzik post a wide ranging survey as an early effort to try to understand, orient, enhance and benefit from this imminent worldwise, computational transition. But the historic occasion of some cerebral, machine neural deep learning, cognizance going on by itself is such a revolutionary presence with many issues and quandaries. The second quote might give some idea. Thus we repurpose, expand and rename an Earthificial Intelligence: Deep Neural Network Computation Planetary Science section. See also Powerful ”Machine Scientists” Distill the Laws of Physics from Raw Data by Charlie Wood in Quanta (May 10, 2022) for another array of novel paths.

Imagine an “oracle” that predicts the outcome of a particle physics experiment, the products of a chemical reaction, or the function of every protein. As scientists, we would not be satisfied, for we need to comprehend how these predictions were conceived. This feat of scientific understanding, has long been the essential aim of science. Now, the ever-growing power of computers and AI poses this question: But today we ask how can advanced computer systems contribute to learning and discovery. At this early phase we seek advice from the philosophy of science, review the state of the art, and ask current researchers about how they acquired novel findings this way. We hope our perspective inspires and focuses research towards devices and methods that foster and empower this worldwide facility. (Abstract excerpt, edit)

Three Dimensions of Computer-Assisted Understanding: We use scientific literature and personal anecdotes of many active users, and the philosophy of science, to introduce a new classification of android contributions to scientific understanding. Such entities can act I) as a computational microscope, providing information not (yet) attainable by experiment, II) as a resource of inspiration or artificial muse,. In those two classes, the human investigator is essential to develop new insights to their full potential Finally, an android can be III) an agent of understanding by generalizing observations and finding novel scientific concepts. (4)

Marcus, Gary. The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence. arXiv:2002.06177. The NYU polypsychologist and founder of Robust AI has rightly situated himself as a reality checker and quality control moderator as this hyper-active endeavor moves into every aspect that it can. See also his Rebooting AI: Building Artificial Intelligence We Can Trust 2019 book.

Recent research in artificial intelligence and machine learning has largely emphasized general-purpose learning and ever-larger training sets and more and more compute. In contrast, I propose a hybrid, knowledge-driven, reasoning-based approach, centered around cognitive models, that could provide the substrate for a richer, more robust AI than is currently possible.

Ornes, Stephen. Researchers Turn to Deep Learning to Decode Protein Structures. PNAS. 119/10, 2022. We note this science report to highlight now the growing broad avail of frontier neural net capabilities that are serving to revolutionize biochemical research and knowledge.

AlphaFold (DeepMind) uses AI to predict the shapes of proteins; structural biologists are using the program to deepen our understanding of the big molecules. This image shows AlphaFold's predicted structure (in magenta) of a glycoprotein found on the surface of a T cell. (1) The revolution in structural biology isn’t attributable to AI alone; the algorithms have to train on big datasets of high-resolution structures generated by technologies such as X-ray crystallography, NMR spectroscopy or cryogenic electron microscopy, which produced the above image of a protein complex called β-galactosidase. (3)

In the future, researchers see a role for deep learning not only in understanding a protein’s shape but also how it interacts within a living system. Deep learning models may predict not only the sequence of amino acids that would produce the needed shape, but also how they’ll behave — and interact with other molecules in their biological neighborhood — once they’re in place. (4)

Park, Sang Eon, et al. Quasi Anomalous Knowledge: Searching for New Physics with Embedded Knowledge. arXiv:2011.03550. This entry by MIT nuclear physicists is an example of how neural net machine methods can advance sub-atomic particle research.

Discoveries of new phenomena often involve a dedicated search for a hypothetical physics signature. Recently, novel deep learning techniques have emerged to detect anomalies but need more precision. Here we present a new strategy dubbed Quasi Anomalous Knowledge (QUAK) which can capture some of the salient features of physics signatures, allowing for the recovery of sensitivity even when signals vary. In this paper, we apply QUAK to anomaly detection of new physics events at the CERN Large Hadron Collider.

Seif, Alireza, et al. Machine Learning the Thermodynamic Arrow of Time. Nature Physics. 17/1, 2022. We cite this entry by University of Maryland physicists including Chris Jarzynski as example of how these 2020s bio-based, neural net techniques which run iterative programmed computations can serve as an advanced spiral stage of worldwise scientific studies. In this case, the old arrow of time problem gains a new depth of understanding which was heretofore inaccessible.

The asymmetry in the flow of events that is expressed as “time’s arrow’” traces back to the second law of thermodynamics. In the microscopic regime, fluctuations prevent us from discerning the direction of time with certainty. Here, we find that a machine learning algorithm trained to infer an actual aim identifies entropy production as the relevant physical quantity in its decision-making process. The algorithm rediscovers the fluctuation theorem as the prime thermodynamic principle. Our results indicate that machine learning methods can be used to study out of equilibrium systems and begin to uncover deep physical principles. (Abstract)

Wood, Charlie. Powerful ”Machine Scientists” Distill the Laws of Physics from Raw Data. Quanta. May 10, 2022. A science writer deftly realizes that a historic shift is underway as current AI methods begin to empower a revolutionary advance in how research studies can be carried out. The report goes on to survey diverse instances with an exemplary focus on the work of Spanish cell biologists Marta Sales-Pardo and Roger Guimera. The entry brings together so many uses that it has led me to rename and expand this Science Spiral section. In regard, we will be posting a plethora of papers from cosmological to quantum domains, and all in between. Wood’s entry begins with late 20th century starts so on to an array of algorithms by which can handle the vast data inputs that now flood in.

For instance, NYU climate physicist Laure Zanna models ocean turbulence, and Flatiron Institute and University of Washington programmers are busy ginning up equations. At Columbia U., Hod Lipson’s group finds agile software for better neural net performance, while Max Tegmark at MIT adds other versions (search both herein). For Sales-Pardo and Guimera’s cellular avail see Regulation of Cell Cycle Progression by Marina Utoz, et al in Nature Cell Biology (20/646, 2018) and A Bayesian Machine Scientist to Aid in the Solution of Challenging Scientific Problems by R. Guimera, et al in Scientific Advances (6/5, 202).

Closed-form, interpretable mathematical models have been instrumental for advancing our understanding of the world. With the data revolution, we may now be in a position to uncover new models for many systems from physics to the social sciences. However, to deal with increasing amounts of data, we need “machine scientists” that are able to extract these models automatically from data. Here, we introduce a Bayesian machine scientist, which considers the plausibility of models by iteratively learning from a large empirical corpus of mathematical expressions. We show that this approach uncovers accurate models for synthetic and real data and provides predictions that are more accurate than other nonparametric methods. (Excerpt)

Wood, Charlie. How to Make the Universe Think for Us.. Quanta. June 1, 2022. The insightful science writer follows up his Powerful “Machine Scientists” Distill the Laws of Physics article (May 10, search) by noting more ways that this Earthificial phase is tailoring deep neural network algorithms so they can seek out, process data, reiterate and come up with findings. A lead citation is Deep Physical Neural Networks with Backpropagation by Logan Wright, et al (MIT, Cornell) in Nature (601/550, 2022), followed by Ben Scellier (Zurich) whose collegial entry is Agnostic Physics-Driven Deep Learning at arXiv:2205.15021 (Abstracts below). Contributions by Julie Grollier, Florian Marquardt, Sam Dillavou (Decentralized, Physics-Driven Learning at 2108.00275) and others are noted as this Earth-Human-Earth mission of universal edification goes forward.

Physicists are building neural networks out of vibrations, voltages and lasers, arguing that the future of computing lies in exploiting the universe’s complex physical behaviors. (CW)

Our approach allows us to train deep physical neural networks made from controllable physical systems, even when the layers lack a mathematical isomorphism to conventional artificial neural networks. To show the universality of our approach, we train diverse physical neural networks based on optics, mechanics and electronics to perform audio and image classification tasks. Physics-aware training combines the scalability of backpropagation with the automatic mitigation of imperfections and noise achievable with in situ algorithms. (Wright)

This work establishes that a physical system can perform statistical learnin via an Agnostic Equilibrium Propagation procedure that combines energy minimization, homeostatic control, and nudging towards a correct response. The procedure is based only on external manipulations, and produces a stochastic gradient descent without explicit gradient computations. This method considerably widens the range of potential hardware for statistical learning to any system with enough controllable parameters. (Scellier)

Yang, Vicky and Anders Sandberg. Collective Intelligence as Infrastructure for Reducing Global Catastrophic Risks. arXiv:2205.03300. Santa Fe Institute and Oxford University, Future of Humanity Institute scholars open this posting by noting how all manner of groupings from cerebral function, fish schools, wildebeest herds, onto political elections can be seen as a natural propensity to form viable associations. By extension, it would be to our advantage if this process could be intentionally availed to serve our own survivability. See also Adaptive Social Networks Promote the Wisdom of Crowds by Almaatouq, A., et al in PNAS (117/21, 2020).

Academic and philanthropic endeavors have grown concerned with global catastrophic risks (GCRs) such as artificial intelligence safety, pandemics, biosecurity, and nuclear war. Study and resolution efforts often depend on the performance of human meetings, which can be seen to involve Collective Intelligence (CI) agendas. CI is a transdisciplinary perspective, whose application involves animal groups, economic markets, collections of neurons, and other distributed systems. Here we argue that better CI methods can improve general resilience against a wide variety of risks. GCR researchers can benefit from engaging more with behavioral researchers to impact critical social issues by engaging and enhancing these transdisciplinary efforts. (Abstract excerpt)

Ecosmos: A Revolutionary Fertile, Habitable, Solar-Bioplanet Lifescape

Overbye, Dennis. The Milky Way’s Black Hole Comes to Light.. New York Times. May 13, 2022. A science reporter lauds this discovery by the Event Horizon Telescope Collaboration team, which was announced last week. Several articles about it in the are accessible from this note. Once again, how fantastic is it that one minute newly sentient personsphere can yet achieve such galactic quantifications. Why does this phenomenon exist at all anyway?

The Event Horizon Telescope is a large telescope array consisting of a global network of radio telescopes. The EHT project combines data from several very-long-baseline interferometry (VLBI) stations around Earth, which form a combined array with an angular resolution sufficient to observe objects the size of a supermassive black hole's event horizon.

Previous   1 | 2 | 3 | 4 | 5 | 6  Next