Controlling Potential Information to Unify Internal Structure for Improved Interpretation and Generalization in Multi-layered Neural Networks

2021 ◽  
pp. 567-588
Author(s):  
Ryotaro Kamimura
Keyword(s):  
Neural Networks ◽  
Internal Structure

scholarly journals Fully interpretable deep learning model of transcriptional control

2020 ◽  
Vol 36 (Supplement_1) ◽  
pp. i499-i507 ◽  
Author(s):  
Yi Liu ◽  
Kenneth Barr ◽  
John Reinitz
Keyword(s):  
Neural Networks ◽  
Systems Biology ◽  
Internal Structure ◽  
Transcriptional Control ◽  
Mathematical Structure ◽  
Fruit Fly ◽  
Early Embryo ◽  
Supplementary Information ◽  
Test Bed ◽  
Specific Chemical

Abstract Motivation The universal expressibility assumption of Deep Neural Networks (DNNs) is the key motivation behind recent worksin the systems biology community to employDNNs to solve important problems in functional genomics and moleculargenetics. Typically, such investigations have taken a ‘black box’ approach in which the internal structure of themodel used is set purely by machine learning considerations with little consideration of representing the internalstructure of the biological system by the mathematical structure of the DNN. DNNs have not yet been applied to thedetailed modeling of transcriptional control in which mRNA production is controlled by the binding of specific transcriptionfactors to DNA, in part because such models are in part formulated in terms of specific chemical equationsthat appear different in form from those used in neural networks. Results In this paper, we give an example of a DNN whichcan model the detailed control of transcription in a precise and predictive manner. Its internal structure is fully interpretableand is faithful to underlying chemistry of transcription factor binding to DNA. We derive our DNN from asystems biology model that was not previously recognized as having a DNN structure. Although we apply our DNNto data from the early embryo of the fruit fly Drosophila, this system serves as a test bed for analysis of much larger datasets obtained by systems biology studies on a genomic scale. . Availability and implementation The implementation and data for the models used in this paper are in a zip file in the supplementary material. Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals Giant Steps In The Interpretation Of A Musical PDP Network

Eureka ◽  
2014 ◽  
Vol 4 (1) ◽  
pp. 24-29
Author(s):  
Joshua Hathaway ◽  
Michael R. W. Dawson
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Internal Structure ◽  
Musical Cognition ◽  
New Directions ◽  
Musical Formalism

We first introduce the notion of chord progressions by describing a particular example (the II-V-I) that is related to the Coltrane changes. Second, we describe the Coltrane changes using a formalism derived from previous musical investigations with neural networks (Yaremchuk & Dawson, 2005, 2008). Finally, we describe how we trained a neural network to generate the Coltrane changes, how we analyzed its internal structure, and the implications of this interpretation. In particular, we discovered that a network represented transitions between chords in a fashion that could be described in terms of a new musical formalism that we had not envisioned. In short, this paper shows that the interpretation of the internal structure of a musical network can provide new formalisms for representing musical regularities, and can suggest new directions for representational research on musical cognition. 

scholarly journals Using deep neural networks to compute the mass of forming planets

2019 ◽  
Vol 626 ◽  
pp. A21 ◽  
Author(s):  
Y. Alibert ◽  
J. Venturini
Keyword(s):  
Neural Networks ◽  
Boundary Conditions ◽  
Internal Structure ◽  
Deep Neural Networks ◽  
Critical Mass ◽  
Support Vector ◽  
Core Mass ◽  
Analytical Formulas ◽  
Structure Equations ◽  
The Difference

Context. Computing the mass of planetary envelopes and the critical mass beyond which planets accrete gas in a runaway fashion is important for studying planet formation, in particular, for planets up to the Neptune-mass range. This computation in principle requires solving a set of differential equations, the internal structure equations, for some boundary conditions (pressure, temperature in the protoplanetary disc where a planet forms, core mass, and the rate of accretion of solids by the planet). Solving these equations in turn proves to be time-consuming and sometimes numerically unstable. Aims. The aim is to provide a way to approximate the result of integrating the internal structure equations for a variety of boundary conditions. Methods. We computed a set of internal planetary structures for a very large number (millions) of boundary conditions, considering two opacities: that of the interstellar medium, and a reduced opacity. This database was then used to train deep neural networks (DNN) in order to predict the critical core mass and the mass of planetary envelopes as a function of the boundary conditions. Results. We show that our neural networks provide a very good approximation (at the percent level) of the result obtained by solving interior structure equations, but the required computer time is much shorter. The difference with the real solution is much smaller than the difference that is obtained with some analytical formulas that are available in the literature, which only provide the correct order of magnitude at best. We compare the results of the DNN with other popular machine-learning methods (random forest, gradient boost, support vector regression) and show that the DNN outperforms these methods by a factor of at least two. Conclusions. We show that some analytical formulas that can be found in various papers can severely overestimate the mass of planets and therefore predict the formation of planets in the Jupiter-mass regime instead of the Neptune-mass regime. The python tools that we provide allow computing the critical mass and the mass of planetary envelopes in a variety of cases, without the requirement of solving the internal structure equations. These tools can easily replace previous analytical formulas and provide far more accurate results.

scholarly journals Increasing the Reliability of Flood Embankments with Neural Imaging Method

2018 ◽  
Vol 8 (9) ◽  
pp. 1457 ◽  
Author(s):  
Grzegorz Kłosowski ◽  
Tomasz Rymarczyk ◽  
Arkadiusz Gola
Keyword(s):  
Neural Networks ◽  
Internal Structure ◽  
Single Point ◽  
Tomographic Reconstruction ◽  
Imaging Method ◽  
Neural Imaging ◽  
Non Invasive ◽  
Innovative System ◽  
High Resolution Images ◽  
Resolution Problem

This paper presents an innovative system of many artificial neural networks that enables the tomographic reconstruction of the internal structure of a flood embankment. An advantage of the proposed method is that it allows us to obtain high-resolution images, which essentially contributes to early, precise and reliable prediction of operational hazards. The method consists in training a cluster of separate neural networks, each of which generates a single point of the output image. The simultaneous and parallel application of the set of neural networks led to effective reconstruction of the internal structure of a deposition site for floatation tailings. Results obtained from the study allow us to solve the low resolution problem that usually occurs with non-invasive imaging methods. This effect was possible thanks to the design of a new intelligent image reconstruction system.

X-ray microtomography

1988 ◽  
Vol 46 ◽  
pp. 998-999
Author(s):  
H.W. Deckman ◽  
B.F. Flannery ◽  
J.H. Dunsmuir ◽  
K.D' Amico
Keyword(s):  
Computed Tomography ◽  
Internal Structure ◽  
Imaging Technique ◽  
Three Dimensional ◽  
Tissue Structure ◽  
Individual Element ◽  
X Ray ◽  
Non Invasive ◽  
Panoramic Imaging ◽  
Three Dimensional Images

We have developed a new X-ray microscope which produces complete three dimensional images of samples. The microscope operates by performing X-ray tomography with unprecedented resolution. Tomography is a non-invasive imaging technique that creates maps of the internal structure of samples from measurement of the attenuation of penetrating radiation. As conventionally practiced in medical Computed Tomography (CT), radiologists produce maps of bone and tissue structure in several planar sections that reveal features with 1mm resolution and 1% contrast. Microtomography extends the capability of CT in several ways. First, the resolution which approaches one micron, is one thousand times higher than that of the medical CT. Second, our approach acquires and analyses the data in a panoramic imaging format that directly produces three-dimensional maps in a series of contiguous stacked planes. Typical maps available today consist of three hundred planar sections each containing 512x512 pixels. Finally, and perhaps of most import scientifically, microtomography using a synchrotron X-ray source, allows us to generate maps of individual element.

The internal structure of medullated wool

1984 ◽  
Vol 42 ◽  
pp. 388-389
Author(s):  
Leo Barish
Keyword(s):  
Light Microscopy ◽  
Internal Structure ◽  
Wool Fiber ◽  
Guard Hair ◽  
Bright Field ◽  
Wool Fibers ◽  
Thin Skin ◽  
Air Pockets

Although most of the wool used today consists of fine, unmedullated down-type fibers, a great deal of coarse wool is used for carpets, tweeds, industrial fabrics, etc. Besides the obvious diameter difference, coarse wool fibers are often medullated.Medullation may be easily observed using bright field light microscopy. Fig. 1A shows a typical fine diameter nonmedullated wool fiber, Fig. IB illustrates a coarse fiber with a large medulla. The opacity of the medulla is due to the inability of the mounting media to penetrate to the center of the fiber leaving air pockets. Fig. 1C shows an even thicker fiber with a very large medulla and with very thin skin. This type of wool is called “Kemp”, is shed annually or more often, and corresponds to guard hair in fur-bearing animals.

Determination of S-N curves with the application of artificial neural networks

1999 ◽  
Vol 22 (8) ◽  
pp. 723-728 ◽  
Author(s):  
Artymiak ◽  
Bukowski ◽  
Feliks ◽  
Narberhaus ◽  
Zenner
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Artificial Neural

The German Post-Critical Belief Scale: Internal and External Validity

2003 ◽  
Vol 34 (4) ◽  
pp. 219-226 ◽  
Author(s):  
Bart Duriez ◽  
Claudia Appel ◽  
Dirk Hutsebaut
Keyword(s):  
Internal Structure ◽  
Present Article ◽  
External Validity ◽  
German Version ◽  
Dutch Version ◽  
German Sample ◽  
Belief Scale ◽  
Internal And External Validity

Abstract: Recently, Duriez, Fontaine and Hutsebaut (2000) and Fontaine, Duriez, Luyten and Hutsebaut (2003) constructed the Post-Critical Belief Scale in order to measure the two religiosity dimensions along which Wulff (1991 , 1997 ) summarized the various possible approaches to religion: Exclusion vs. Inclusion of Transcendence and Literal vs. Symbolic. In the present article, the German version of this scale is presented. Results obtained in a heterogeneous German sample (N = 216) suggest that the internal structure of the German version fits the internal structure of the original Dutch version. Moreover, the observed relation between the Literal vs. Symbolic dimension and racism, which was in line with previous studies ( Duriez, in press ), supports the external validity of the German version.

Internal Structure of the Spanish Adaptation of the Eating Disorder Inventory-3

2012 ◽  
Vol 28 (1) ◽  
pp. 25-31 ◽  
Author(s):  
Paula Elosua ◽  
Alicia López-Jáuregui
Keyword(s):  
Eating Disorders ◽  
Eating Disorder ◽  
Internal Structure ◽  
Internal Consistency ◽  
Clinical Sample ◽  
Temporal Stability ◽  
Second Order ◽  
Order Structure ◽  
Eating Disorder Inventory ◽  
High Internal Consistency

In this study the Eating Disorder Inventory-3 was adapted to Spanish and analyzed the internal psychometric properties of the test in a clinical sample of females with eating disorders. The results showed a high internal consistency of the scores as well as high temporal stability. The factor structure of the scale composites was analyzed using confirmatory factor analysis. The results supported the existence of a second-order structure beyond the psychological composites. The second-order factor showed high correlation with the factor related to eating disorders. Overall, the Spanish version of the EDI-3 showed good psychometric qualities in terms of internal consistency, temporal stability and internal structure.

Sign in / Sign up

Export Citation Format

Share Document