scholarly journals Learning Without Feedback: Fixed Random Learning Signals Allow for Feedforward Training of Deep Neural Networks

2021 ◽  
Vol 15 ◽  
Author(s):  
Charlotte Frenkel ◽  
Martin Lefebvre ◽  
David Bol
Keyword(s):  
Supervised Classification ◽  
Low Cost ◽  
Computational Cost ◽  
Random Projections ◽  
Smart Sensors ◽  
Classification Problems ◽  
Network Training ◽  
Memory Accesses ◽  
The One ◽  
Target Projection

While the backpropagation of error algorithm enables deep neural network training, it implies (i) bidirectional synaptic weight transport and (ii) update locking until the forward and backward passes are completed. Not only do these constraints preclude biological plausibility, but they also hinder the development of low-cost adaptive smart sensors at the edge, as they severely constrain memory accesses and entail buffering overhead. In this work, we show that the one-hot-encoded labels provided in supervised classification problems, denoted as targets, can be viewed as a proxy for the error sign. Therefore, their fixed random projections enable a layerwise feedforward training of the hidden layers, thus solving the weight transport and update locking problems while relaxing the computational and memory requirements. Based on these observations, we propose the direct random target projection (DRTP) algorithm and demonstrate that it provides a tradeoff between accuracy and computational cost that is suitable for adaptive edge computing devices.

Drug Target Group Prediction with Multiple Drug Networks

2020 ◽  
Vol 23 (4) ◽  
pp. 274-284 ◽  
Author(s):  
Jingang Che ◽  
Lei Chen ◽  
Zi-Han Guo ◽  
Shuaiqun Wang ◽  
Aorigele
Keyword(s):  
Drug Target ◽  
Low Cost ◽  
Machine Learning Algorithms ◽  
Classification Model ◽  
Support Vector ◽  
Multiple Drug ◽  
Property A ◽  
Multiple Networks ◽  
Proposed Model ◽  
The One

Background: Identification of drug-target interaction is essential in drug discovery. It is beneficial to predict unexpected therapeutic or adverse side effects of drugs. To date, several computational methods have been proposed to predict drug-target interactions because they are prompt and low-cost compared with traditional wet experiments. Methods: In this study, we investigated this problem in a different way. According to KEGG, drugs were classified into several groups based on their target proteins. A multi-label classification model was presented to assign drugs into correct target groups. To make full use of the known drug properties, five networks were constructed, each of which represented drug associations in one property. A powerful network embedding method, Mashup, was adopted to extract drug features from above-mentioned networks, based on which several machine learning algorithms, including RAndom k-labELsets (RAKEL) algorithm, Label Powerset (LP) algorithm and Support Vector Machine (SVM), were used to build the classification model. Results and Conclusion: Tenfold cross-validation yielded the accuracy of 0.839, exact match of 0.816 and hamming loss of 0.037, indicating good performance of the model. The contribution of each network was also analyzed. Furthermore, the network model with multiple networks was found to be superior to the one with a single network and classic model, indicating the superiority of the proposed model.

scholarly journals Analytical Model of Induction Machines with Multiple Cage Faults Using the Winding Tensor Approach

Sensors ◽  
2021 ◽  
Vol 21 (15) ◽  
pp. 5076
Author(s):  
Javier Martinez-Roman ◽  
Ruben Puche-Panadero ◽  
Angel Sapena-Bano ◽  
Carla Terron-Santiago ◽  
Jordi Burriel-Valencia ◽  
...  
Keyword(s):  
Computational Cost ◽  
Induction Machines ◽  
Diagnostic Techniques ◽  
Analytical Models ◽  
Tensor Algebra ◽  
Diagnostic Systems ◽  
Technical Literature ◽  
Twin Model ◽  
Ring Segment ◽  
The One

Induction machines (IMs) are one of the main sources of mechanical power in many industrial processes, especially squirrel cage IMs (SCIMs), due to their robustness and reliability. Their sudden stoppage due to undetected faults may cause costly production breakdowns. One of the most frequent types of faults are cage faults (bar and end ring segment breakages), especially in motors that directly drive high-inertia loads (such as fans), in motors with frequent starts and stops, and in case of poorly manufactured cage windings. A continuous monitoring of IMs is needed to reduce this risk, integrated in plant-wide condition based maintenance (CBM) systems. Diverse diagnostic techniques have been proposed in the technical literature, either data-based, detecting fault-characteristic perturbations in the data collected from the IM, and model-based, observing the differences between the data collected from the actual IM and from its digital twin model. In both cases, fast and accurate IM models are needed to develop and optimize the fault diagnosis techniques. On the one hand, the finite elements approach can provide highly accurate models, but its computational cost and processing requirements are very high to be used in on-line fault diagnostic systems. On the other hand, analytical models can be much faster, but they can be very complex in case of highly asymmetrical machines, such as IMs with multiple cage faults. In this work, a new method is proposed for the analytical modelling of IMs with asymmetrical cage windings using a tensor based approach, which greatly reduces this complexity by applying routine tensor algebra to obtain the parameters of the faulty IM model from the healthy one. This winding tensor approach is explained theoretically and validated with the diagnosis of a commercial IM with multiple cage faults.

scholarly journals The S66 Non-Covalent Interactions Benchmark Reconsidered Using Explicitly Correlated Methods Near the Basis Set Limit

2018 ◽  
Vol 71 (4) ◽  
pp. 238 ◽  
Author(s):  
Manoj K. Kesharwani ◽  
Amir Karton ◽  
Nitai Sylvetsky ◽  
Jan M. L. Martin
Keyword(s):  
Computational Cost ◽  
The Other ◽  
Basis Set ◽  
Basis Sets ◽  
Explicitly Correlated ◽  
Non Covalent Interactions ◽  
Explicitly Correlated Methods ◽  
The One ◽  
High Level ◽  
Covalent Interactions

The S66 benchmark for non-covalent interactions has been re-evaluated using explicitly correlated methods with basis sets near the one-particle basis set limit. It is found that post-MP2 ‘high-level corrections’ are treated adequately well using a combination of CCSD(F12*) with (aug-)cc-pVTZ-F12 basis sets on the one hand, and (T) extrapolated from conventional CCSD(T)/heavy-aug-cc-pV{D,T}Z on the other hand. Implications for earlier benchmarks on the larger S66×8 problem set in particular, and for accurate calculations on non-covalent interactions in general, are discussed. At a slight cost in accuracy, (T) can be considerably accelerated by using sano-V{D,T}Z+ basis sets, whereas half-counterpoise CCSD(F12*)(T)/cc-pVDZ-F12 offers the best compromise between accuracy and computational cost.

scholarly journals Status Estimation and In-Process Connection of Kanbans Using BLE Beacons and LPWA Network to Implement Intra-Traceability for the Kanban System

Sensors ◽  
2021 ◽  
Vol 21 (15) ◽  
pp. 5038
Author(s):  
Kosuke Shima ◽  
Masahiro Yamaguchi ◽  
Takumi Yoshida ◽  
Takanobu Otsuka
Keyword(s):  
Inventory Management ◽  
Low Cost ◽  
Computational Cost ◽  
Low Energy ◽  
Manufacturing Strategy ◽  
Implementation Cost ◽  
Measurement Systems ◽  
Process Route ◽  
Kanban System ◽  
Low Computational Cost

IoT-based measurement systems for manufacturing have been widely implemented. As components that can be implemented at low cost, BLE beacons have been used in several systems developed in previous research. In this work, we focus on the Kanban system, which is a measure used in manufacturing strategy. The Kanban system emphasizes inventory management and is used to produce only required amounts. In the Kanban system, the Kanban cards are rotated through the factory along with the products, and when the products change to a different process route, the Kanban card is removed from the products and the products are assigned to another Kanban. For this reason, a single Kanban cannot trace products from plan to completion. In this work, we propose a system that uses a Bluetooth low energy (BLE) beacon to connect Kanbans in different routes but assigned to the same products. The proposed method estimates the beacon status of whether the Kanban is inside or outside a postbox, which can then be computed by a micro controller at low computational cost. In addition, the system connects the Kanbans using the beacons as paired connection targets. In an experiment, we confirmed that the system connected 70% of the beacons accurately. We also confirmed that the system could connect the Kanbans at a small implementation cost.

Rapid and Green One-Pot Synthesis of Novel 2-(4-Hydroxy-2-oxo-2H-chromen-3-yl)-2-(arylamino)-1H-indene-1,3(2H)-diones

2021 ◽  
Vol 18 ◽  
Author(s):  
Abolfazl Olyaei ◽  
Zahra Ghahremany ◽  
Madieh Sadeghpour
Keyword(s):  
Aromatic Amines ◽  
Low Cost ◽  
Guanidine Hydrochloride ◽  
Reaction Times ◽  
Solvent Free ◽  
One Pot ◽  
Solvent Free Conditions ◽  
The One ◽  
High Yields ◽  
Work Up

: A green and efficient protocol was developed for the one-pot three-component synthesis of novel 2-(4-hydroxy-2-oxo-2H-chromen-3-yl)-2-(arylamino)-1H-indene-1,3(2H)-dione derivatives by the reaction of 4-hydroxycoumarin, ninhydrin and aromatic amines in the presence of guanidine hydrochloride as an organocatalyst under solvent-free conditions. The present approach offers several advantages such as low cost, simple work-up, short reaction times, chromatography-free purification, high yields and greener conditions.

Design of a Directional Olfactory Display to Study the Integration of Vision and Olfaction

2018 ◽  
Author(s):  
Lorenzo Micaroni ◽  
Marina Carulli ◽  
Francesco Ferrise ◽  
Monica Bordegoni ◽  
Alberto Gallace
Keyword(s):  
Virtual Reality ◽  
Low Cost ◽  
Olfactory Cues ◽  
Input Device ◽  
Experimental Setup ◽  
Initial Testing ◽  
Head Mounted Display ◽  
Innovative System ◽  
The One ◽  
Olfactory Display

This research aims to design and develop an innovative system, based on an olfactory display, to be used for investigating the directionality of the sense of olfaction. In particular, the design of an experimental setup to understand and determine to what extent the sense of olfaction is directional and whether there is prevalence of the sense of vision over the one of smell when determining the direction of an odor, is described. The experimental setup is based on low cost Virtual Reality (VR) technologies. In particular, the system is based on a custom directional olfactory display, an Oculus Rift Head Mounted Display (HMD) to deliver both visual and olfactory cues and an input device to register subjects’ answers. The VR environment is developed in Unity3D. The paper describes the design of the olfactory interface as well as its integration with the overall system. Finally the results of the initial testing are reported in the paper.

scholarly journals Teaching and Learning Physics with Smartphones

2015 ◽  
Vol 17 (1) ◽  
pp. 31-50 ◽  
Author(s):  
M. Á. González ◽  
Manuel Á. González ◽  
M. Esther Martín ◽  
César Llamas ◽  
Óscar Martínez ◽  
...  
Keyword(s):  
Mobile Applications ◽  
Teaching And Learning ◽  
Low Cost ◽  
Positive Influence ◽  
Mobile Technologies ◽  
Traditional Learning ◽  
Know How ◽  
Measurement Devices ◽  
The One ◽  
Physics Experiments

The use of mobile technologies is reshaping how to teach and learn. In this paper the authors describe their research on the use of these technologies to teach physics. On the one hand they develop mobile applications to complement the traditional learning and to help students learn anytime and anywhere. The use of this applications has proved to have very positive influence on the students' engagement. On the other hand, they use smartphones as measurement devices in physics experiments. This opens the possibility of designing and developing low cost laboratories where expensive material can be substituted by smartphones. The smartphones' sensors are reliable and accurate enough to permit good measurements. However, as it is shown with some examples, special care must be taken here if one does not know how these apps used to access the sensors' data are programmed.

Identification of suitable nontraditional tropical and residual paving materials in relation to the environment

1987 ◽  
Vol 24 (1) ◽  
pp. 58-71 ◽  
Author(s):  
M. D. Gidigasu ◽  
S. P. K. Asante ◽  
E. Dougan
Keyword(s):  
Performance Studies ◽  
Low Cost ◽  
Index Properties ◽  
Moisture Index ◽  
Fines Content ◽  
Geotechnical Characteristics ◽  
Pavement Construction ◽  
The One ◽  
Identification And Characterization

Results of laboratory and field construction and pavement performance studies have revealed that some apparently substandard or borderline—by temperature zone standards—tropical gravel materials can be used successfully for the construction of low-cost pavements in subhumid climatic zones.The geotechnical characteristics of suitable paving materials have been defined from results of studies carried out on satisfactory and poor road sections over a period of 5–6 years.Important correlations were found between the bearing strength (CBR) on the one hand and some moisture index properties as well as the products of the fines content and these index properties on the other.It is shown that the product of the fines content and some moisture index properties form a useful basis for identifying suitable materials for pavement construction in the moist subhumid climatic zone of Ghana.The paper is a contribution to the identification and characterization of nontraditional tropical and residual paving materials in relation to the environment. Key words: pavements, California bearing ratio (CBR), index properties, fines content, subhumid climate.

scholarly journals Binary and Multiclass Text Classification by Means of Separable Convolutional Neural Network

Inventions ◽  
2021 ◽  
Vol 6 (4) ◽  
pp. 70
Author(s):  
Elena Solovyeva ◽  
Ali Abdullah
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Convolutional Neural Network ◽  
Recurrent Neural Networks ◽  
Low Cost ◽  
Computational Cost ◽  
High Accuracy ◽  
Activation Functions ◽  
Fully Connected ◽  
Fully Connected Networks

In this paper, the structure of a separable convolutional neural network that consists of an embedding layer, separable convolutional layers, convolutional layer and global average pooling is represented for binary and multiclass text classifications. The advantage of the proposed structure is the absence of multiple fully connected layers, which is used to increase the classification accuracy but raises the computational cost. The combination of low-cost separable convolutional layers and a convolutional layer is proposed to gain high accuracy and, simultaneously, to reduce the complexity of neural classifiers. Advantages are demonstrated at binary and multiclass classifications of written texts by means of the proposed networks under the sigmoid and Softmax activation functions in convolutional layer. At binary and multiclass classifications, the accuracy obtained by separable convolutional neural networks is higher in comparison with some investigated types of recurrent neural networks and fully connected networks.

scholarly journals More and Faster: Simultaneously Improving Reaction Coverage and Computational Cost in Automated Reaction Prediction Tasks

2020 ◽  
Author(s):  
Qiyuan Zhao ◽  
Brett Savoie
Keyword(s):  
Transition State ◽  
Domain Knowledge ◽  
Low Cost ◽  
Computational Cost ◽  
Chemical Degradation ◽  
Reaction Pathways ◽  
Reaction Networks ◽  
Complex Reaction ◽  
Success Rates ◽  
Reaction Prediction

<div> <div> <div> <p>Automated reaction prediction has the potential to elucidate complex reaction networks for applications ranging from combustion to materials degradation. Although substantial progress has been made in predicting specific reaction pathways and resolving mechanisms, the computational cost and inconsistent reaction coverage of automated prediction are still obstacles to exploring deep reaction networks without using heuristics. Here we show that cost can be reduced and reaction coverage can be increased simultaneously by relatively straight- forward modifications of the reaction enumeration, geometry initialization, and transition state convergence algorithms that are common to many emerging prediction methodologies. These changes are implemented in the context of Yet Another Reaction Program (YARP), our reaction prediction package, for which we report a head-to-head comparison with prevailing methods for two benchmark reaction prediction tasks. In all cases, we observe near perfect recapitulation of established reaction pathways and products by YARP, without the use of heuristics or other domain knowledge to guide reaction selection. In addition, YARP also discovers many new kinetically relevant pathways and products reported here for the first time. This is achieved while simultaneously reducing the cost of reaction characterization nearly 100-fold and increasing transition state success rates and intended rates over 2-fold and 10-fold, respectively, compared with recent benchmarks. This combination of ultra-low cost and high reaction-coverage creates opportunities to explore the reactivity of larger sys- tems and more complex reaction networks for applications like chemical degradation, where approaches based on domain heuristics fail. </p> </div> </div> </div>

Sign in / Sign up

Export Citation Format

Share Document