1,720,961 research outputs found
AI3SD Video: Interpretable Machine Learning for Materials’ Design and Characterisation
No full text“Where is the knowledge we have lost in information?” T.S. Eliot, The RockMachine learning (ML) and artificial intelligence (AI) are the subjects of wildly differing opinions on utility and potential impact. Depending who we talk to ML is the solution to almost every human challenge, from open boarders to pandemic control, or presents an existential crises for the species. In materials science the polarisation is perhaps less extreme, but nonetheless pervasive, while the numbers of ML related works experiences an explosion one highly respected theoretical chemist recently pronounced “[a]t least 50% of the machine learning papers I see regarding electronic structure are junk”. A part of the issue that many detractors have with ML methods is related to their perception of the techniques as ‘black-box’ approaches, at the same time, the same lack of understanding limitations of the models leads to some of the more outlandish boosterism surrounding the subject. In this talk I will discuss, with examples from our work, how we can open up the black-box of ML methods, highlighting and understanding limitations, increasing trust in results, and potentially improving the methods themselves
AI3SD Project: Artificial intelligence for reconstruction and super-resolution of chemical tomography
No full textX-ray scatter-based tomography allows unprecedented insight into the chemical and physical state of functional materials and devices. Such tomographies can be used as research tools but also offer the prospect of routine scanning for security and inspection systems and potential for medical scanning. However X-ray scatter tomogrpahy requires longer collection times and higher doses than conventional absorption tomography - in this project we will develop machine learning tools to fuse X-ray scatter and X-ray absorption tomogrpahy, providing the detail ofthe former with the efficiency of the latter.In conventional X-ray tomography, the images that are obtained give maps of density withinthe object and the composing pixels contain single grey scale values. In scatter based tomography, each pixel instead contains spectrum or equivalent chemical signal i.e. a 1D array (or higher) of numbers. An X-ray scatter tomography slice becomes a data cube with the two conventional spatial dimensions and a third spectral dimension. Such image data is termedhyperspectral
AI3SD Video: Calibrated deep representations and entropy based active learning for materials property prediction
No full textThis talk forms part of the ML4MC (Machine Learning for Materials and Chemicals Series which has been organised as a joint venture between the Artificial Intelligence for Scientific Discovery Network+ (AI3SD) and the Directed Assembly Network. This series ran over summer 2021 and covers topics that encompass our overlapping Network interests of AI, Machine Learning, Artificial Photosynthesis, Biomimetic Materials, Crystal Design & Engineering, Materials, Molecules, Photochemistry, Photocatalysis and Supramolecular Chemistry. This video was the ninth talk in the ML4MC series and formed part of the session "Mentor Talks"
AI3SD Video: Interpretable machine learning for materials design and characterization
No full textIn a plenary lecture at a recent international conference, one leading researcher in theoretical chemistry remarked "at least 50% of the machine learning papers I see regarding electronic structure theory are junk, and do not meet the minimal standards of scientific publication", specifically referring to the lack of insight in many publications applying ML in that field. But is knowledge inevitably lost in machine learning studies, if not how can it be extracted and how does this apply to machine learning in the context of materials science? In this talk I will look at how we can open up black box machine learning models, to understand the results and gain confidence in predictions. I will present topical examples from designing new dielectric crystals, understanding inelastic neutron scattering data and trusting deep neural networks for tomographic reconstruction. By understanding how and why these models work, we can trust the results and even discover new physical relationships
Humans of AI3SD: Dr Keith Butler
No full textThis interview forms part of our Humans of AI3SD Series. Keith Butler is a Senior Data Scientist working on materials science research in the SciML team at Rutherford Appleton Laboratory. SciML is a team in the Scientific Computing Division working with the large STFC facilities (Diamond, ISIS Neutron and Muon Source and Central Laser Facility, for example) to use machine learning to push the boundaries of fundamental science. In this Humans of AI3SD interview he discusses the impact of his work, the potential of self- driving labs, the importance of explainable and interpretable machine learning systems and why early career researchers should shout about what they know (and use Linux!)
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Full text link“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Full text linkWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Dispelling the Myths Behind First-author Citation Counts
Full text linkWe conducted a full-scale evaluative citation analysis study of scholars in the XML research field to explore just how different from each other author rankings resulting from different citation counting methods actually are, and to demonstrate the capability of emerging data and tools on the Web in supporting more realistic citation counting methods. Our results contest some common arguments for the continued
use of first-author citation counts in the evaluation of scholars, such as high correlations between author rankings by first-author citation counts and other citation
counting methods, and high costs of using more realistic citation counting methods that are not well-supported by the ISI databases. It is argued that increasingly available digital full text research papers make it possible for citation analysis studies to go beyond what the ISI databases have directly supported and to employ more
sophisticated methods
- …
