Search results for: graph convolution network

Authors: Guanghua Zhang, Fubao Wang, Weijun Duan

Abstract:

Convolution neural network (CNN) has attracted more and more attention on recent years. Especially in the field of computer vision and image classification. However, unsupervised learning with CNN has received less attention than supervised learning. In this work, we use a new powerful tool which is deep convolutional generative adversarial networks (DCGANs) to generate images from Sloan Digital Sky Survey. Training by various star and galaxy images, it shows that both the generator and the discriminator are good for unsupervised learning. In this paper, we also took several experiments to choose the best value for hyper-parameters and which could help to stabilize the training process and promise a good quality of the output.

Keywords: convolution neural network, discriminator, generator, unsupervised learning

Procedia PDF Downloads 265

Authors: Qiuyi Lyu, Bin Gong

Abstract:

Breadth-First Search(BFS) is a core graph algorithm that is widely used for graph analysis. As it is frequently used in many graph applications, improve the BFS performance is essential. In this paper, we present a graph ordering method that could reorder the graph nodes to achieve better data locality, thus, improving the BFS performance. Our method is based on an observation that the sibling relationships will dominate the cache access pattern during the BFS traversal. Therefore, we propose a frequency-based model to construct the graph order. First, we optimize the graph order according to the nodes’ visit frequency. Nodes with high visit frequency will be processed in priority. Second, we try to maximize the child nodes overlap layer by layer. As it is proved to be NP-hard, we propose a heuristic method that could greatly reduce the preprocessing overheads. We conduct extensive experiments on 16 real-world datasets. The result shows that our method could achieve comparable performance with the state-of-the-art methods while the graph ordering overheads are only about 1/15.

Keywords: breadth-first search, BFS, graph ordering, graph algorithm

Procedia PDF Downloads 136

Authors: Gowtham Kalkere Jayanna, Mohamad Nazri Husin

Abstract:

Graph theory, chemistry, and technology are all combined in cheminformatics. The structure and physiochemical properties of organic substances are linked using some useful graph invariants and the corresponding molecular graph. In this paper, we study specific reverse topological indices such as the reverse sum-connectivity index, the reverse Zagreb index, the reverse arithmetic-geometric, and the geometric-arithmetic, the reverse Sombor, the reverse Nirmala indices for the bistar graphs B (n: m) and the corona product Kₘ∘Kₙ', where Kₙ' Represent the complement of a complete graph Kₙ.

Keywords: reverse topological indices, bistar graph, the corona product, graph

Procedia PDF Downloads 94

Authors: Lin Cheng, Zijiang Yang

Abstract:

Program synthesis is the task to automatically generate programs based on user specification. In this paper, we present a framework that synthesizes programs from flow charts that serve as accurate and intuitive specification. In order doing so, we propose a deep neural network called GRCNN that recognizes graph structure from its image. GRCNN is trained end-to-end, which can predict edge and node information of the flow chart simultaneously. Experiments show that the accuracy rate to synthesize a program is 66.4%, and the accuracy rates to recognize edge and node are 94.1% and 67.9%, respectively. On average, it takes about 60 milliseconds to synthesize a program.

Keywords: program synthesis, flow chart, specification, graph recognition, CNN

Procedia PDF Downloads 118

Authors: Gaye Ibrahima, Mendy Gervais, Seck Diaraf, Ouya Samuel

Abstract:

In this paper, we consider the influence maximization in social networks. Here we give importance to initial diffuser called the seeds. The goal is to find efficiently a subset of k elements in the social network that will begin and maximize the information diffusion process. A new approach which treats the social network before to determine the seeds, is proposed. This treatment eliminates the information feedback toward a considered element as seed by extracting an acyclic spanning social network. At first, we propose two algorithm versions called SCG − algoritm (v1 and v2) (Spanning Connected Graphalgorithm). This algorithm takes as input data a connected social network directed or no. And finally, a generalization of the SCG − algoritm is proposed. It is called SG − algoritm (Spanning Graph-algorithm) and takes as input data any graph. These two algorithms are effective and have each one a polynomial complexity. To show the pertinence of our approach, two seeds set are determined and those given by our approach give a better results. The performances of this approach are very perceptible through the simulation carried out by the R software and the igraph package.

Keywords: acyclic spanning graph, centrality measures, information feedback, influence maximization, social network

Procedia PDF Downloads 248

Authors: S. R. Nayaka, Putta Swamy

Abstract:

Graph polynomial is one of the algebraic representations of the Graph. The degree polynomial is one of the simple algebraic representations of graphs. The degree polynomial of a graph G of order n is the polynomial Deg(G, x) with the coefficients deg(G,i) where deg(G,i) denotes the number of vertices of degree i in G. In this article, we investigate the behavior of the roots of some families of Graphs in the complex field. We investigate for the graphs having only integral roots. Further, we characterize the graphs having single roots or having real roots and behavior of the polynomial at the particular value is also obtained.

Keywords: degree polynomial, regular graph, minimum and maximum degree, graph operations

Procedia PDF Downloads 246

Authors: Cao Xiaopeng, Shi Linkai

Abstract:

The practical Byzantine fault-tolerant algorithm does not add nodes dynamically. It is limited in practical application. In order to add nodes dynamically, Dynamic Practical Byzantine Fault Tolerance Algorithm (DPBFT) was proposed. Firstly, a new node sends request information to other nodes in the network. The nodes in the network decide their identities and requests. Then the nodes in the network reverse connect to the new node and send block information of the current network. The new node updates information. Finally, the new node participates in the next round of consensus, changes the view and selects the master node. This paper abstracts the decision of nodes into the undirected connected graph. The final consistency of the graph is used to prove that the proposed algorithm can adapt to the network dynamically. Compared with the PBFT algorithm, DPBFT has better fault tolerance and lower network bandwidth.

Keywords: practical byzantine, fault tolerance, blockchain, consensus algorithm, consistency analysis

Procedia PDF Downloads 128

Authors: Dimitra Alexiou

Abstract:

In an urban area the allocation placement of an emergency service mobile units, such as ambulances, police patrol must be designed so as to achieve a prompt response to demand locations. In this paper, a partition of a given urban network into distinct sub-networks is performed such that; the vertices in each component are close and simultaneously the difference of the sums of the corresponding population in the sub-networks is almost uniform. The objective here is to position appropriately in each sub-network a mobile emergency unit in order to reduce the response time to the demands. A mathematical model in the framework of graph theory is developed. In order to clarify the corresponding method a relevant numerical example is presented on a small network.

Keywords: graph partition, emergency service, distances, location

Procedia PDF Downloads 498

Authors: Andrey Lebedev, Ilya Dmitrenok, JooYoung Lee, Leonard Johard

Abstract:

Top-k shortest path routing problem is an extension of finding the shortest path in a given network. Shortest path is one of the most essential measures as it reveals the relations between two nodes in a network. However, in many real world networks, whose diameters are small, top-k shortest path is more interesting as it contains more information about the network topology. Many variations to compute top-k shortest paths have been studied. In this paper, we apply an efficient top-k shortest distance routing algorithm to the link prediction problem and test its efficacy. We compare the results with other base line and state-of-the-art methods as well as with the shortest path. Then, we also propose a top-k distance based graph matching algorithm.

Keywords: graph matching, link prediction, shortest path, similarity

Procedia PDF Downloads 356

Authors: Ladznar S. Laja, Rosalio G. Artes, Jr.

Abstract:

This paper introduced a graph polynomial relating convexity concepts. A graph polynomial is a polynomial representing a graph given some parameters. On the other hand, a subgraph H of a graph G is said to be convex in G if for every pair of vertices in H, every shortest path with these end-vertices lies entirely in H. We define the convex subgraph polynomial of a graph G to be the generating function of the sequence of the numbers of convex subgraphs of G of cardinalities ranging from zero to the order of G. This graph polynomial is monic since G itself is convex. The convex index which counts the number of convex subgraphs of G of all orders is just the evaluation of this polynomial at 1. Relationships relating algebraic properties of convex subgraphs polynomial with graph theoretic concepts are established.

Keywords: convex subgraph, convex index, generating function, polynomial ring

Procedia PDF Downloads 213

Authors: Raghavi C. Janaswamy

Abstract:

In the healthcare sector, heterogenous data elements such as patients, diagnosis, symptoms, conditions, observation text from physician notes, and prescriptions form the essentials of the Electronic Health Record (EHR). The data in the form of clear text and images are stored or processed in a relational format in most systems. However, the intrinsic structure restrictions and complex joins of relational databases limit the widespread utility. In this regard, the design and development of realistic mapping and deep connections as real-time objects offer unparallel advantages. Herein, a graph neural network-based classification of EHR data has been developed. The patient conditions have been predicted as a node classification task using a graph-based open source EHR data, Synthea Database, stored in Tigergraph. The Synthea DB dataset is leveraged due to its closer representation of the real-time data and being voluminous. The graph model is built from the EHR heterogeneous data using python modules, namely, pyTigerGraph to get nodes and edges from the Tigergraph database, PyTorch to tensorize the nodes and edges, PyTorch-Geometric (PyG) to train the Graph Neural Network (GNN) and adopt the self-supervised learning techniques with the AutoEncoders to generate the node embeddings and eventually perform the node classifications using the node embeddings. The model predicts patient conditions ranging from common to rare situations. The outcome is deemed to open up opportunities for data querying toward better predictions and accuracy.

Keywords: electronic health record, graph neural network, heterogeneous data, prediction

Procedia PDF Downloads 85

Authors: Vadim E. Levit, Eugen Mandrescu

Abstract:

A graph is well-covered if all its maximal independent sets are of the same size. A well-covered graph is 1-well-covered if deletion of every vertex of the graph leaves it well-covered. It is known that a graph without isolated vertices is 1-well-covered if and only if every two disjoint independent sets are included in two disjoint maximum independent sets. Well-covered graphs are related to combinatorial commutative algebra (e.g., every Cohen-Macaulay graph is well-covered, while each Gorenstein graph without isolated vertices is 1-well-covered). Our intent is to construct several infinite families of 1-well-covered graphs using the following known graph operations: corona, join, and rooted product of graphs. Adopting some known techniques used to advantage for well-covered graphs, one can prove that: if the graph G has no isolated vertices, then the corona of G and H is 1-well-covered if and only if H is a complete graph of order two at least; the join of the graphs G and H is 1-well-covered if and only if G and H have the same independence number and both are 1-well-covered; if H satisfies the property that every three pairwise disjoint independent sets are included in three pairwise disjoint maximum independent sets, then the rooted product of G and H is 1-well-covered, for every graph G. These findings show not only how to generate some more families of 1-well-covered graphs, but also that, to this aim, sometimes, one may use graphs that are not necessarily 1-well-covered.

Keywords: maximum independent set, corona, concatenation, join, well-covered graph

Procedia PDF Downloads 207

Authors: Chi Wu, Dean Wu, Wen-Te Liu, Cheng-Yu Tsai, Shin-Mei Hsu, Yin-Tzu Lin, Ru-Yin Yang

Abstract:

Background: The results of previous studies compared the benefits and risks of receiving insomnia medication. However, the effects between hypnotic drugs used and enhancement of sleep quality were still unclear. Objective: The aim of this study is to establish a prediction model for hypnotic drugs' dosage used for insomnia subjects and associated the relationship between sleep stage ratio change and drug types. Methodologies: According to American Academy of Sleep Medicine (AASM) guideline, sleep stages were classified and transformed to hypnogram via the polysomnography (PSG) in a hospital in New Taipei City (Taiwan). The subjects with diagnosis for insomnia without receiving hypnotic drugs treatment were be set as the comparison group. Conversely, hypnotic drugs dosage within the past three months was obtained from the clinical registration for each subject. Furthermore, the collecting subjects were divided into two groups for training and testing. After training convolution neuron network (CNN) to predict types of hypnotics used and dosages are taken, the test group was used to evaluate the accuracy of classification. Results: We recruited 76 subjects in this study, who had been done PSG for transforming hypnogram from their sleep stages. The accuracy of dosages obtained from confusion matrix on the test group by CNN is 81.94%, and accuracy of hypnotic drug types used is 74.22%. Moreover, the subjects with high ratio of wake stage were correctly classified as requiring medical treatment. Conclusion: CNN with hypnogram was potentially used for adjusting the dosage of hypnotic drugs and providing subjects to pre-screening the types of hypnotic drugs taken.

Keywords: convolution neuron network, hypnotic drugs, insomnia, polysomnography

Procedia PDF Downloads 194

Authors: Khidir R. Sharaf, Didar A. Ali

Abstract:

The nullity η (G) of a graph is the occurrence of zero as an eigenvalue in its spectra. A zero-sum weighting of a graph G is real valued function, say f from vertices of G to the set of real numbers, provided that for each vertex of G the summation of the weights f (w) over all neighborhood w of v is zero for each v in G.A high zero-sum weighting of G is one that uses maximum number of non-zero independent variables. If G is graph with an end vertex, and if H is an induced sub-graph of G obtained by deleting this vertex together with the vertex adjacent to it, then, η(G)= η(H). In this paper, a high zero-sum weighting technique and the end vertex procedure are applied to evaluate the nullity of t-tupple and generalized t-tupple graphs are derived and determined for some special types of graphs. Also, we introduce and prove some important results about the t-tupple coalescence, Cartesian and Kronecker products of nut graphs.

Keywords: graph theory, graph spectra, nullity of graphs, statistic

Procedia PDF Downloads 238

Authors: Nivedha Rajaram

Abstract:

Hybrid Quantum solvers have been given prime focus in recent days by computation problem-solving domain industrial applications. D’Wave Quantum Computers are one such paragon of systems built using quantum annealing mechanism. Discrete Quadratic Models is a hybrid quantum computing model class supplied by D’Wave Ocean SDK - a real-time software platform for hybrid quantum solvers. These hybrid quantum computing modellers can be employed to solve classic problems. One such problem that we consider in this paper is finding a network connectivity knowledge hub in a huge network of systems. Using this quantum solver, we try to find out the prime system hub, which acts as a supreme connection point for the set of connected computers in a large network. This paper establishes an innovative problem approach to generate a connectivity system hub plot for a set of systems using DWave ocean SDK hybrid quantum solvers.

Keywords: quantum computing, hybrid quantum solver, DWave annealing, network knowledge graph

Procedia PDF Downloads 125

Authors: Jimmy Eadie

Abstract:

Convolutional Cross Synthesis (CCS) has emerged as a powerful technique for blending input signals to create hybrid sounds. It has significantly expanded the horizons of digital signal processing, enabling artists to explore audio effects. However, the conventional applications of CCS primarily revolve around reverberation and room simulation rather than being utilized as a creative synthesis method. In this paper, we present the design of a digital instrument called CrossSampler that harnesses a parametric approach to convolution cross-synthesis, which involves using adjustable parameters to control the blending of audio signals through convolution. These parameters allow for customization of the resulting sound, offering greater creative control and flexibility. It enables users to shape the output by manipulating factors such as duration, intensity, and spectral characteristics. This approach facilitates experimentation and exploration in sound design and opens new sonic possibilities.

Keywords: convolution, synthesis, sampling, virtual instrument

Procedia PDF Downloads 62

Authors: Margarita Mayoral-Villa, J. Klapp, L. Di G. Sigalotti, J. E. V. Guzmán

Abstract:

Automated learning techniques are widely applied in the energy sector to address challenging problems from a practical point of view. To this end, we discuss the implementation of a Message Passing algorithm (MPNN)within a Graph Neural Network(GNN)to leverage the neighborhood of a set of nodes during the aggregation process. This approach enables the characterization of multiphase diffusion processes in the reservoir, such that the flow paths underlying the interconnections between multiple wells may be inferred from previously available data on flow rates and bottomhole pressures. The results thus obtained compare favorably with the predictions produced by the Reduced Order Capacitance-Resistance Models (CRM) and suggest the potential of MPNNs to enhance the robustness of the forecasts while improving the computational efficiency.

Keywords: multiphase diffusion, message passing neural network, well interconnection, interwell connectivity, graph neural network, capacitance-resistance models

Procedia PDF Downloads 147

Authors: Nirmal Yadav, Tanuja Srivastava

Abstract:

Shannon theory is an exact method to recover a band limited signals from its sampled values in discrete implementation, using sinc interpolators. But sinc based results are not much satisfactory for band-limited calculations so that convolution with window function, having compact support, has been introduced. Convolution Backprojection algorithm with window function is an approximation algorithm. In this paper, the error has been calculated, arises due to this approximation nature of reconstruction algorithm. This result will be defined for fan beam projection data which is more faster than parallel beam projection.

Keywords: computed tomography, convolution backprojection, radon transform, fan beam

Procedia PDF Downloads 490

Authors: Krishnendra Shekhawat

Abstract:

Given a graph G = (V, E), a rectangular dual of G represents the vertices of G by a set of interior-disjoint rectangles such that two rectangles touch if and only if there is an edge between the two corresponding vertices in G. Rectangular duals do not exist for every graph, so we can define maximal rectangular duals. A maximal rectangular dual is a rectangular dual of a graph G such that there exists no graph G ′ with a rectangular dual where G is a subgraph of G ′. In this paper, we enumerate all maximal rectangular duals (or, to be precise, the corresponding planar graphs) up to six nodes and presents a necessary condition for the existence of a rectangular dual. This work allegedly has applications in integrated circuit design and architectural floor plans.

Keywords: adjacency, degree sequence, dual graph, rectangular dual

Procedia PDF Downloads 265

Authors: Pallavi Bhogaram, Xiaolong Wu, Min He, Onyedikachi Okenwa

Abstract:

A transportation network is a realization of a spatial network, describing a structure which permits either vehicular movement or flow of some commodity. Examples include road networks, railways, air routes, pipelines, and many more. The transportation network plays a vital role in maintaining the vigor of the nation’s economy. Hence, ensuring the network stays resilient all the time, especially in the face of challenges such as heavy traffic loads and large scale natural disasters, is of utmost importance. In this paper, we used the Neo4j application to develop the graph. Neo4j is the world's leading open-source, NoSQL, a native graph database that implements an ACID-compliant transactional backend to applications. The Southern California network model is developed using the Neo4j application and obtained the most critical and optimal nodes and paths in the network using centrality algorithms. The edge betweenness centrality algorithm calculates the critical or optimal paths using Yen's k-shortest paths algorithm, and the node betweenness centrality algorithm calculates the amount of influence a node has over the network. The preliminary study results confirm that the Neo4j application can be a suitable tool to study the important nodes and the critical paths for the major congested metropolitan area.

Keywords: critical path, transportation network, connectivity reliability, network model, Neo4j application, edge betweenness centrality index

Procedia PDF Downloads 133

Authors: Asraful Islam

Abstract:

Extended dependency graph (EDG) is a state-of-the-art isomorphic graph to represent normal logic programs (NLPs) that can characterize the consistency of NLPs by graph analysis. To construct the vertices and arcs of an EDG, additional renaming atoms and rules besides those the given program provides are used, resulting in higher space complexity compared to the corresponding traditional dependency graph (TDG). In this article, we propose an extended labeled dependency graph (ELDG) to represent an NLP that shares an equal number of nodes and arcs with TDG and prove that it is isomorphic to the domain program. The number of nodes and arcs used in the underlying dependency graphs are formulated to compare the space complexity. Results show that ELDG uses less memory to store nodes, arcs, and cycles compared to EDG. To exhibit the desirability of ELDG, firstly, the stable models of the kernel form of NLP are characterized by the admissible coloring of ELDG; secondly, a relation of the stable models of a kernel program with the handles of the minimal, odd cycles appearing in the corresponding ELDG has been established; thirdly, to our best knowledge, for the first time an inverse transformation from a dependency graph to the representing NLP w.r.t. ELDG has been defined that enables transferring analytical results from the graph to the program straightforwardly.

Keywords: normal logic program, isomorphism of graph, extended labelled dependency graph, inverse graph transforma-tion, graph colouring

Procedia PDF Downloads 210

Authors: Ashutosh Mishra, Nikhil Goyal

Abstract:

A 3D facial emotion recognition model based on deep learning is proposed in this paper. Two convolution layers and a pooling layer are employed in the deep learning architecture. After the convolution process, the pooling is finished. The probabilities for various classes of human faces are calculated using the sigmoid activation function. To verify the efficiency of deep learning-based systems, a set of faces. The Kaggle dataset is used to verify the accuracy of a deep learning-based face recognition model. The model's accuracy is about 65 percent, which is lower than that of other facial expression recognition techniques. Despite significant gains in representation precision due to the nonlinearity of profound image representations.

Keywords: facial recognition, computational intelligence, convolutional neural network, depth map

Procedia PDF Downloads 229

Authors: Zahra Ramezanpanah, Joachim Carvallo, Aurelien Rodriguez

Abstract:

This paper aims to detect anomalies in streaming data using machine learning algorithms. In this regard, we designed two separate pipelines and evaluated the effectiveness of each separately. The first pipeline, based on supervised machine learning methods, consists of two phases. In the first phase, we trained several supervised models using the UNSW-NB15 data-set. We measured the efficiency of each using different performance metrics and selected the best model for the second phase. At the beginning of the second phase, we first, using Argus Server, sniffed a local area network. Several types of attacks were simulated and then sent the sniffed data to a running algorithm at short intervals. This algorithm can display the results of each packet of received data in real-time using the trained model. The second pipeline presented in this paper is based on unsupervised algorithms, in which a Temporal Graph Network (TGN) is used to monitor a local network. The TGN is trained to predict the probability of future states of the network based on its past behavior. Our contribution in this section is introducing an indicator to identify anomalies from these predicted probabilities.

Keywords: temporal graph network, anomaly detection, cyber security, IDS

Procedia PDF Downloads 102

Authors: Po-Jen Chen, Jian-Jiun Ding, Hung-Wei Hsu, Chien-Yao Wang, Jia-Ching Wang

Abstract:

A more accurate scene classification algorithm using label relations and the hierarchy neural network was developed in this work. In many classification algorithms, it is assumed that the labels are mutually exclusive. This assumption is true in some specific problems, however, for scene classification, the assumption is not reasonable. Because there are a variety of objects with a photo image, it is more practical to assign multiple labels for an image. In this paper, two label relations, which are exclusive relation and hierarchical relation, were adopted in the classification process to achieve more accurate multiple label classification results. Moreover, the hierarchy neural network (hierarchy NN) is applied to classify the image and the directed acyclic graph structure is used for predicting a more reasonable result which obey exclusive and hierarchical relations. Simulations show that, with these techniques, a much more accurate scene classification result can be achieved.

Keywords: convolutional neural network, label relation, hierarchy neural network, scene classification

Procedia PDF Downloads 455

Authors: Jaeyoung Lee

Abstract:

Autonomous driving systems require high reliability to provide people with a safe and comfortable driving experience. However, despite the development of a number of vehicle sensors, it is difficult to always provide high perceived performance in driving environments that vary from time to season. The image segmentation method using deep learning, which has recently evolved rapidly, provides high recognition performance in various road environments stably. However, since the system controls a vehicle in real time, a highly complex deep learning network cannot be used due to time and memory constraints. Moreover, efficient networks are optimized for GPU environments, which degrade performance in embedded processor environments equipped simple hardware accelerators. In this paper, a semantic segmentation network, matrix multiplication accelerator network (MMANet), optimized for matrix multiplication accelerator (MMA) on Texas instrument digital signal processors (TI DSP) is proposed to improve the recognition performance of autonomous driving system. The proposed method is designed to maximize the number of layers that can be performed in a limited time to provide reliable driving environment information in real time. First, the number of channels in the activation map is fixed to fit the structure of MMA. By increasing the number of parallel branches, the lack of information caused by fixing the number of channels is resolved. Second, an efficient convolution is selected depending on the size of the activation. Since MMA is a fixed, it may be more efficient for normal convolution than depthwise separable convolution depending on memory access overhead. Thus, a convolution type is decided according to output stride to increase network depth. In addition, memory access time is minimized by processing operations only in L3 cache. Lastly, reliable contexts are extracted using the extended atrous spatial pyramid pooling (ASPP). The suggested method gets stable features from an extended path by increasing the kernel size and accessing consecutive data. In addition, it consists of two ASPPs to obtain high quality contexts using the restored shape without global average pooling paths since the layer uses MMA as a simple adder. To verify the proposed method, an experiment is conducted using perfsim, a timing simulator, and the Cityscapes validation sets. The proposed network can process an image with 640 x 480 resolution for 6.67 ms, so six cameras can be used to identify the surroundings of the vehicle as 20 frame per second (FPS). In addition, it achieves 73.1% mean intersection over union (mIoU) which is the highest recognition rate among embedded networks on the Cityscapes validation set.

Keywords: edge network, embedded network, MMA, matrix multiplication accelerator, semantic segmentation network

Procedia PDF Downloads 128

Authors: Ja-Keoung Koo, Kensuke Nakamura, Hyohun Kim, Dongwha Shin, Yeonseok Kim, Ji-Su Ahn, Byung-Woo Hong

Abstract:

The machine learning techniques based on a convolutional neural network (CNN) have been actively developed and successfully applied to a variety of image analysis tasks including reconstruction, noise reduction, resolution enhancement, segmentation, motion estimation, object recognition. The classical visual information processing that ranges from low level tasks to high level ones has been widely developed in the deep learning framework. It is generally considered as a challenging problem to derive visual interpretation from high dimensional imagery data. A CNN is a class of feed-forward artificial neural network that usually consists of deep layers the connections of which are established by a series of non-linear operations. The CNN architecture is known to be shift invariant due to its shared weights and translation invariance characteristics. However, it is often computationally intractable to optimize the network in particular with a large number of convolution layers due to a large number of unknowns to be optimized with respect to the training set that is generally required to be large enough to effectively generalize the model under consideration. It is also necessary to limit the size of convolution kernels due to the computational expense despite of the recent development of effective parallel processing machinery, which leads to the use of the constantly small size of the convolution kernels throughout the deep CNN architecture. However, it is often desired to consider different scales in the analysis of visual features at different layers in the network. Thus, we propose a CNN model where different sizes of the convolution kernels are applied at each layer based on the random projection. We apply random filters with varying sizes and associate the filter responses with scalar weights that correspond to the standard deviation of the random filters. We are allowed to use large number of random filters with the cost of one scalar unknown for each filter. The computational cost in the back-propagation procedure does not increase with the larger size of the filters even though the additional computational cost is required in the computation of convolution in the feed-forward procedure. The use of random kernels with varying sizes allows to effectively analyze image features at multiple scales leading to a better generalization. The robustness and effectiveness of the proposed CNN based on random kernels are demonstrated by numerical experiments where the quantitative comparison of the well-known CNN architectures and our models that simply replace the convolution kernels with the random filters is performed. The experimental results indicate that our model achieves better performance with less number of unknown weights. The proposed algorithm has a high potential in the application of a variety of visual tasks based on the CNN framework. Acknowledgement—This work was supported by the MISP (Ministry of Science and ICT), Korea, under the National Program for Excellence in SW (20170001000011001) supervised by IITP, and NRF-2014R1A2A1A11051941, NRF2017R1A2B4006023.

Keywords: deep learning, convolutional neural network, random kernel, random projection, dimensionality reduction, object recognition

Procedia PDF Downloads 285

Authors: Hassan Ibrahim

Abstract:

The prevalence of hyperacusis, an auditory condition characterized by heightened sensitivity to sounds, continues to rise, posing challenges for effective diagnosis and intervention. It is believed that this work deepens will deepens the understanding of hyperacusis etiology by employing graph theory as a novel analytical framework. We constructed a comprehensive graph wherein nodes represent various factors associated with hyperacusis, including aging, head or neck trauma, infection/virus, depression, migraines, ear infection, anxiety, and other potential contributors. Relationships between factors are modeled as edges, allowing us to visualize and quantify the interactions within the etiological landscape of hyperacusis. it employ the concept of the metric dimension of a connected graph to identify key nodes (landmarks) that serve as critical influencers in the interconnected web of hyperacusis causes. This approach offers a unique perspective on the relative importance and centrality of different factors, shedding light on the complex interplay between physiological, psychological, and environmental determinants. Visualization techniques were also employed to enhance the interpretation and facilitate the identification of the central nodes. This research contributes to the growing body of knowledge surrounding hyperacusis by offering a network-centric perspective on its multifaceted causes. The outcomes hold the potential to inform clinical practices, guiding healthcare professionals in prioritizing interventions and personalized treatment plans based on the identified landmarks within the etiological network. Through the integration of graph theory into hyperacusis research, the complexity of this auditory condition was unraveled and pave the way for more effective approaches to its management.

Keywords: auditory condition, connected graph, hyperacusis, metric dimension

Procedia PDF Downloads 38

Authors: Hassan Ibrahim

Abstract:

The prevalence of hyperacusis, an auditory condition characterized by heightened sensitivity to sounds, continues to rise, posing challenges for effective diagnosis and intervention. It is believed that this work deepens will deepens the understanding of hyperacusis etiology by employing graph theory as a novel analytical framework. it constructed a comprehensive graph wherein nodes represent various factors associated with hyperacusis, including aging, head or neck trauma, infection/virus, depression, migraines, ear infection, anxiety, and other potential contributors. Relationships between factors are modeled as edges, allowing us to visualize and quantify the interactions within the etiological landscape of hyperacusis. it employ the concept of the metric dimension of a connected graph to identify key nodes (landmarks) that serve as critical influencers in the interconnected web of hyperacusis causes. This approach offers a unique perspective on the relative importance and centrality of different factors, shedding light on the complex interplay between physiological, psychological, and environmental determinants. Visualization techniques were also employed to enhance the interpretation and facilitate the identification of the central nodes. This research contributes to the growing body of knowledge surrounding hyperacusis by offering a network-centric perspective on its multifaceted causes. The outcomes hold the potential to inform clinical practices, guiding healthcare professionals in prioritizing interventions and personalized treatment plans based on the identified landmarks within the etiological network. Through the integration of graph theory into hyperacusis research, the complexity of this auditory condition was unraveled and pave the way for more effective approaches to its management.

Keywords: auditory condition, connected graph, hyperacusis, metric dimension

Procedia PDF Downloads 19

Authors: Qingqing Gai, Junxing Shen, Yu Luo

Abstract:

Due to the hardware architecture characteristics, some operators tend to acquire better performance if the input/output tensor dimensions are aligned to a certain minimum granularity, such as convolution and deconvolution commonly used in deep learning. Furthermore, if the requirements are not met, the general strategy is to pad with 0 to satisfy the requirements, potentially leading to the under-utilization of the hardware resources. Therefore, for the convolution and deconvolution whose input and output channels do not meet the minimum granularity alignment, we propose to transfer the W-dimensional data to the C-dimension for computation (W2C) to enable the C-dimension to meet the hardware requirements. This scheme also reduces the number of computations in the W-dimension. Although this scheme substantially increases computation, the operator’s speed can improve significantly. It achieves remarkable speedups on multiple hardware accelerators, including Nvidia Tensor cores, Qualcomm digital signal processors (DSPs), and Huawei neural processing units (NPUs). All you need to do is modify the network structure and rearrange the operator weights offline without retraining. At the same time, for some operators, such as the Reducemax, we observe that transferring the Cdimensional data to the W-dimension(C2W) and replacing the Reducemax with the Maxpool can accomplish acceleration under certain circumstances.

Keywords: convolution, deconvolution, W2C, C2W, alignment, hardware accelerator

Procedia PDF Downloads 103

Authors: Shibo Wei, Ting Jiang

Abstract:

Convolutional-recurrent neural networks (CRN) have achieved much success recently in the speech enhancement field. The common processing method is to use the convolution layer to compress the feature space by multiple upsampling and then model the compressed features with the LSTM layer. At last, the enhanced speech is obtained by deconvolution operation to integrate the global information of the speech sequence. However, the feature space compression process may cause the loss of information, so we propose to model the upsampling result of each step with the residual LSTM layer, then join it with the output of the deconvolution layer and input them to the next deconvolution layer, by this way, we want to integrate the global information of speech sequence better. The experimental results show the network model (RES-CRN) we introduce can achieve better performance than LSTM without residual and overlaying LSTM simply in the original CRN in terms of scale-invariant signal-to-distortion ratio (SI-SNR), speech quality (PESQ), and intelligibility (STOI).

Keywords: convolutional-recurrent neural networks, speech enhancement, residual LSTM, SI-SNR

Procedia PDF Downloads 198