Filters








433,776 Hits in 7.9 sec

Activation Functions: Comparison of trends in Practice and Research for Deep Learning [article]

Chigozie Nwankpa, Winifred Ijomah, Anthony Gachagan, Stephen Marshall
2018 arXiv   pre-print
To achieve these state-of-the-art performances, the DL architectures use activation functions (AFs), to perform diverse computations between the hidden layers and the output layers of any given DL architecture  ...  the state-of-the-art research results.  ...  TABLE I I TYPES AND POSITIONS OF AFS USED IN DL ARCHITECTURES.recent architecture and understand the architectural make-up of the network.  ... 
arXiv:1811.03378v1 fatcat:qfwc3ywnarhi7gdnzhbcks2mom

Chemception: A Deep Neural Network with Minimal Chemistry Knowledge Matches the Performance of Expert-developed QSAR/QSPR Models [article]

Garrett B. Goh, Charles Siegel, Abhinav Vishnu, Nathan O. Hodas, Nathan Baker
2017 arXiv   pre-print
We then show how Chemception can serve as a general-purpose neural network architecture for predicting toxicity, activity, and solvation properties when trained on a modest database of 600 to 40,000 compounds  ...  Having matched the performance of expert-developed QSAR/QSPR deep learning models, our work demonstrates the plausibility of using deep neural networks to assist in computational chemistry research, where  ...  can serve as a general-purpose neural network for learning a range of distinct properties, including physiological (toxicity), biochemical (activity) and physical (free energy of solvation) properties  ... 
arXiv:1706.06689v1 fatcat:mlm5bvfgsjdlbdg7lu4l4howqe

Using Rule-Based Labels for Weak Supervised Learning: A ChemNet for Transferable Chemical Property Prediction [article]

Garrett B. Goh, Charles Siegel, Abhinav Vishnu, Nathan O. Hodas
2018 arXiv   pre-print
Our results indicate a pre-trained ChemNet that incorporates chemistry domain knowledge, enables the development of generalizable neural networks for more accurate prediction of novel chemical properties  ...  Furthermore, we demonstrate that the ChemNet pre-training approach is equally effective on both CNN (Chemception) and RNN (SMILES2vec) models, indicating that this approach is network architecture agnostic  ...  Nathan Baker for helpful discussions. is work is supported by the following PNNL LDRD programs: Pauling Postdoctoral Fellowship and Deep Learning for Scienti c Discovery Agile Investment.  ... 
arXiv:1712.02734v2 fatcat:itrjobfzkzexnlw5nqwxjqmzk4

Active distributed management for IP networks

R. Kawamura, R. Stadler
2000 IEEE Communications Magazine  
This motivates us to introduce a new management architecture, named Active Distributed Management, which exploits the active network and mobile agent paradigms and provides the properties of distributed  ...  control and programmability inside the network.  ...  In this article we first explain the properties of active technologies which we find necessary for a distributed and programmable management architecture.  ... 
doi:10.1109/35.833567 fatcat:rntr7xpzpzebdkhqvvigdikaqy

IT-based information management in health care networks: the MedoCom approach

Nils Hellrung, Nathalie Gusew, Martin Willkomm, Reinhold Haux
2008 Studies in Health Technology and Informatics  
The framework is used to deduce effects of network characteristics on information management. Concluding, we present an architectural approach of a management platform for health care networks.  ...  Health care networks can be characterized by complex and even conflicting goal systems.  ...  Outlook Figure 1 : 1 Management decisions in centralized and decentralized networks Figure 2 2 depicts the architectural approach of MedoCom and shows the component diagram of MedoCom architecture design  ... 
pmid:18487800 fatcat:poe7yyrzprhirm5yxcpkuku6tm

The EPC Sensor Network for RFID and WSN Integration Infrastructure

Jongwoo Sung, Tomas Sanchez Lopez, Daeyoung Kim
2007 Fifth Annual IEEE International Conference on Pervasive Computing and Communications Workshops (PerComW'07)  
Despite the popularity of current Radio Frequency Identification (RFID) and Wireless Sensor Networks (WSN), current research fails to propose the global vision that is needed for truly pervasive computing  ...  In this paper we introduce our effort to build a global standard infrastructure for WSN and RFID based on the EPCglobal standard Architecture Framework.  ...  There are no standard based research activities which can integrate separate RFID and WSN into one network, probably due to a limited research focus and different research backgrounds.  ... 
doi:10.1109/percomw.2007.113 dblp:conf/percom/SungLK07 fatcat:kmh3dnpfjzfoti675wdrkizgaa

CheMixNet: Mixed DNN Architectures for Predicting Chemical Properties using Multiple Molecular Representations [article]

Arindam Paul, Dipendra Jha, Reda Al-Bahrani, Wei-keng Liao, Alok Choudhary, Ankit Agrawal
2018 arXiv   pre-print
In this work, we present CheMixNet -- a set of neural networks for predicting chemical properties from a mixture of features learned from the two molecular representations -- SMILES as sequences and molecular  ...  The proposed CheMixNet models not only outperforms the candidate neural architectures such as contemporary fully connected networks that uses molecular fingerprints and 1-D CNN and RNN models trained SMILES  ...  Department of Commerce, National Institute of Standards and Technology as part of the Center for Hierarchical Materials Design (CHiMaD).  ... 
arXiv:1811.08283v2 fatcat:zigst4l7szhf5dzophrqmcqt2m

SMILES2Vec: An Interpretable General-Purpose Deep Neural Network for Predicting Chemical Properties [article]

Garrett B. Goh, Nathan O. Hodas, Charles Siegel, Abhinav Vishnu
2018 arXiv   pre-print
Using Bayesian optimization methods to tune the network architecture, we show that an optimized SMILES2vec model can serve as a general-purpose neural network for predicting distinct chemical properties  ...  Our work demonstrates that neural networks can learn technically accurate chemical concept and provide state-of-the-art accuracy, making interpretable deep neural networks a useful tool of relevance to  ...  chemistry is therefore predicated on correlating engineered features with the activity or property of the chemical, which is formally known as the eld of antitative Structure-Activity or Structure-Property  ... 
arXiv:1712.02034v2 fatcat:zgbcrunn7jhqda3inwnohapp5q

Experimentally constrained network model of hippocampal fast-firing parvalbumin-positive interneurons

Katie A Ferguson, Carey YL Huh, Bénédicte Amilhon, Rosanah Murugesu, Sylvain Williams, Frances K Skinner
2012 BMC Neuroscience  
Since the firing properties and network architecture of PV+ interneurons puts them in an ideal position to influence network activity, this cell type will likely remain a focus of experimentalists and  ...  Our network model is composed of these individual PV+ cell models, and the network size, architecture, and synaptic properties are chosen to be consistent with those found in the literature.  ...  Acknowledgements This work was supported by the Canadian Institutes of Health Research and Natural Sciences and Engineering Research Council of Canada.  ... 
doi:10.1186/1471-2202-13-s1-o5 pmcid:PMC3403308 fatcat:f32am7ct5rfujd6rnvbdksig4u

Agile Design of Sustainable Networked Enterprises

Frank M. Lillehagen, John Krogstie
2015 The Practice of Enterprise Modeling  
Use-cases in selected fields have been implemented by agile modelling and holistic design of collaborative networking capabilities, and active knowledge architecture driven solutions.  ...  The active knowledge architecture is the knowledge base for collaborative planning, execution, validation, enhancement and reuse.  ...  The discovery of enterprise knowledge spaces and design modelling of active knowledge architectures of kinds of enterprises has been the main contributions from the use-cases so far implemented.  ... 
dblp:conf/ifip8-1/LillehagenK15 fatcat:g6hfj7ylh5g2df3dmk4ytbzeqm

Neural Networks: Biological Models and Applications [chapter]

F.H. Guenther
2001 International Encyclopedia of the Social & Behavioral Sciences  
properties of individual neurons and networks of interconnected neurons.  ...  At various points in the history of neural network research, successful models have moved beyond the domain of biological modeling into a variety of engineering and medical applications.  ...  Modeling the Computational Properties of Neurons Although the idea that the brain is the seat of the mind and controller of behavior is many centuries old, research into the computational properties of  ... 
doi:10.1016/b0-08-043076-7/03667-6 fatcat:szo3vc37hfguzofztvwohmeywm

Tensile Strength Prediction of Fiberglass Polymer Composites Using Artificial Neural Network Model

Paulina Spanu, Bogdan Felician Abaz
2022 Materiale plastice  
Highlighting the properties of polymer composites is a complex process given their great diversity and the wide range in which their characteristics could vary.  ...  The dependence of the tensile strength on the volume fraction was investigated and highlighted by modelling using neural networks.  ...  In this research it was used a feedforward neural network with sigmoidal unipolar activation function [5] .  ... 
doi:10.37358/mp.22.2.5590 fatcat:zfdneocknrdhxbtuareee2nbay

Building a Semantic Ontology for Internet of Things (IoT) Systems

Allen Ronald DeSerranno, Matthew T. Mullarkey, Alan R. Hevner
2017 Joint Seminar on Ontology Research in Brazil / International Workshop on Metamodels, Ontologies and Semantic Technologies  
The complexity of Internet of Things (IoT) systems requires designers, operators, and users to understand both the structure and semantics embedded in the IoT architecture.  ...  We suggest that a focus on the semantics of the system and the entities in the system be enriched with a definition and understanding of FSQ semantics.  ...  properties and the controls involved in the IoT network (e.g. system structures).  ... 
dblp:conf/ontobras/DeSerrannoMH17 fatcat:zme75kc5gjegnojezc3244bbhu

Nature, nurture, and the development of functional specializations: A computational approach

Robert A. Jacobs
1997 Psychonomic Bulletin & Review  
This principle has been instantiated in a family of neural network architectures referred to as "mixtures-ofexperts" architectures.  ...  Indeed, the notion of modularity motivates significant portions of current research in the cognitive neurosciences, including research on perception, language, motor control, memory, and neural systems  ...  The activation of a unit is computed on the basis of the weighted sum of the activations of the units that project to it.  ... 
doi:10.3758/bf03210788 fatcat:jvwfzk2ucjdjzlv5x6qpz2jtxi

Computationalism, Connectionism, and the Philosophy of Mind [chapter]

Brian P. McLaughlin
2008 The Blackwell Guide to the Philosophy of Computing and Information  
The symbols-system paradigm and the connectionist paradigm are the two dominant research paradigms within the computational theory of mind.  ...  They differ primarily in what kind of computer the mind is assumed to be, and thus in the kinds of functional architectures explored.  ...  Depending on the network architecture, a unit may have only two states of activation, "on" and "off," three or more discrete states of activation, or continuous levels of activation, bounded or unbounded  ... 
doi:10.1002/9780470757017.ch10 fatcat:5chty6rykra5jgz26c2ozvvbgu
« Previous Showing results 1 — 15 out of 433,776 results