DOCSLIB.ORG

Graph Streaming Processor a Next-Generation Computing Architecture

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Graph Streaming Processor a Next-Generation Computing Architecture Graph Streaming Processor A Next-Generation Computing Architecture Val G. Cook – Chief Software Architect Satyaki Koneru – Chief Technology Officer Ke Yin – Chief Scientist Dinakar Munagala – Chief Executive Officer THINCI, Inc. • www.thinci.com August 2017 Introduction • THINCI, Inc. “think-eye” is 5-year-old strategic/venture-backed technology startup • Develop silicon for machine learning, computer vision and other strategic parallel workloads • Provide innovative software along with a comprehensive SDK • 69-person team (95% engineering & operations) • Key IP (patents, trade secrets) – Streaming Graph Processor – Graph Computing Compiler • Product Status – Early Access Program started Q1 2017 – First edition PCIe-based development boards will ship Q4 2017 THINCI, Inc. • www.thinci.com August 2017 Architectural Objective Exceptional efficiency via balanced application of multiple parallel execution mechanisms Levels of Parallelism Key Architectural Choices • Task Level Parallelism • Direct Graph Processing • Thread Level Parallelism • Fine-Grained Thread Scheduling • Data Level Parallelism • 2D Block Processing • Parallel Reduction Instructions • Instruction Level Parallelism • Hardware Instruction Scheduling THINCI, Inc. • www.thinci.com August 2017 Task Level Parallelism Direct Graph Processing THINCI, Inc. • www.thinci.com August 2017 Task Graphs • Formalized Task Level Parallelism – Graphs define only computational semantics – Nodes reference kernels A – Kernels are programs – Nodes bind to buffers – Buffers contain structured data B C – Data dependencies explicit • ThinCI Hardware Processes Graphs Natively – A graph is an execution primitive D E F – A program is a proper sub-set of graph G THINCI, Inc. • www.thinci.com August 2017 Graph Based Frameworks • Graph Processing or Data Flow Graphs – They are a very old concept, for example Alan Turing’s “Graph Turing Machine”. – Gaining value as a computation model, particularly in the field of machine learning. • Graph-based machine learning frameworks have proliferated in recent years. Machine Learning Frameworks TensorFlow Lasagne Chainer CNTK maxDNN MxNet Neural Designer leaf cuDNN Karas DSSTNE Caffe MatConvNet Apache Kaldi Torch BIDMach deeplearning4j SINGA Caffe2 2011 2012 2013 2014 2015 2016 2017 THINCI, Inc. • www.thinci.com August 2017 Streaming vs. Sequential Processing • Sequential Node Processing Sequential Execution Streaming Execution – Commonly used by DSPs and GPUs 0 0 – Intermediate buffers are written back and forth to memory A A – Intermediate buffers are generally 1 2 non-cacheable globally 1 2 – DRAM accesses are costly • Excessive power B C B C • Excessive latency 3 4 5 3 4 5 • Graph Streaming Processor – Intermediate buffers are small D D (~1% of the original size) – Data is more easily cached 6 6 – Benefits of significantly reduced Node A Node B Node C Node D Nodes A,B,C,D memory bandwidth 1 3 5 6 • Lower power consumption 2 4 • Higher performance time time THINCI, Inc. • www.thinci.com August 2017 Thread Level Parallelism Fine-Grained Thread Scheduling THINCI, Inc. • www.thinci.com August 2017 Fine-Grained Thread Scheduling • Thread Scheduler Controller DMA Execution Thread Quad 0 Input/ – Aware of data Command Command Scheduler Special Op Unit Output Ring Unit Ring Unit Unit Processor 0 Processor 1 Processor 2 Processor 3 Thread Thread Thread Thread State State State State dependencies Read Only Inst. Scheduler Inst. Scheduler Inst. Scheduler Inst. Scheduler Cache Array Transfer SPU MPU SPU MPU SPU MPU SPU MPU (DMA) Read Write Read Write Unit Arbiter Cache – Dispatches threads when: Cache Array • Resources available Quad 1 Special Op Unit L2 Cache Processor 0 Processor 1 Processor 2 Processor 3 Thread Thread Thread Thread • Dependencies satisfied State State State State Instruction Inst. Scheduler Inst. Scheduler Inst. Scheduler Inst. Scheduler Unit SPU MPU SPU MPU SPU MPU SPU MPU Read Only – Maintains ordered AXI Bus Matrix Cache Arbiter L3 Cache State behavior as needed Unit Quad 2 Special Op Unit Read Only Cache Processor 0 Processor 1 Processor 2 Processor 3 – Thread Thread Thread Thread Prevents dead-lock State State State State Inst. Scheduler Inst. Scheduler Inst. Scheduler Inst. Scheduler SPU MPU SPU MPU SPU MPU SPU MPU • Supports Complex Scenarios Arbiter – Aggregates Threads Quad N Special Op Unit – Fractures Threads Processor 0 Processor 1 Processor 2 Processor 3 Thread Thread Thread Thread State State State State Inst. Scheduler Inst. Scheduler Inst. Scheduler Inst. Scheduler SPU MPU SPU MPU SPU MPU SPU MPU Arbiter THINCI, Inc. • www.thinci.com August 2017 Graph Execution Trace • Threads can execute from all nodes of the graph simultaneously • True hardware managed streaming behavior Graph Execution Trace Thread life-span Thread Count/Node Thread time THINCI, Inc. • www.thinci.com August 2017 Data Level Parallelism 2D Block Processing Parallel Reduction Instructions THINCI, Inc. • www.thinci.com August 2017 2D Block Processing/Reduction Instructions • Persistent data structures are accessed in blocks • Arbitrary alignment support 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 • Provides for “in-place compute” 1 src • Parallel reduction instructions 2 3 support efficient processing 4 – Reduced power 5 src – Greater throughput 6 7 – Reduced bandwidth 8 9 • Experience better scaling across dst 10 data types vs. the 2x scaling of 11 traditional vector pipelines THINCI, Inc. • www.thinci.com August 2017 Instruction Level Parallelism Hardware Instruction Scheduling THINCI, Inc. • www.thinci.com August 2017 Hardware Instruction Scheduling • Scheduling Groups of Four Processors – Hardware Instruction Picker Thread State Register Files Vector Pipeline – Selects from 100’s of threads Scalar Pipeline – Targets 10’s of independent pipelines Custom Arithmetic Instruction Scheduler Flow Control Instruction Decode Memory Ops. Move Pipeline Thread Spawn State Mgmt. THINCI, Inc. • www.thinci.com August 2017 Programming Model THINCI, Inc. • www.thinci.com August 2017 Programming Model • Fully Programmable – No a-priori constraints regarding data types, precision or graph topologies – Fully pipelined concurrent graph execution – Comprehensive SDK with support for all abstraction levels, assembly to frameworks • Machine Learning Frameworks – TensorFlow – Caffe – Torch • OpenVX + OpenCL C/C++ Language Kernels (Seeking Khronos conformance post Si) – Provides rich graph creation and execution semantics – Extended with fully accelerated custom kernel support THINCI, Inc. • www.thinci.com August 2017 Results • Arithmetic Pipeline Utilization – 95% for CNN’s (VGG16, 8-bit) • Physical Characteristics – TSMC 28nm HPC+ – Standalone SoC Mode – PCIe Accelerator Mode – SoC Power Estimate: 2.5W THINCI, Inc. • www.thinci.com August 2017.
Load more

Recommended publications
  • Comparativa De Tecnologías De Deep Learning
    Universidad Carlos III de Madrid Departamento de Informática Proyecto Final de Carrera Comparativa de Tecnologías de Deep Learning Author: Alberto Lozano Benjumea Supervisors: Alejandro Baldominos Gómez Ignacio Navarro Martín Leganés.Septiembre de 2017 Alberto Lozano Benjumea Universidad Carlos III de Madrid Avenida de la Universidad, 30 28911 Leganés, Madrid (Spain) Email: [email protected] “Just because it’s not nice doesn’t mean it’s not miraculous” Terry Pratchett This page has been intentionally left blank. Abstract Este proyecto se localiza en la rama de aprendizaje automático del campo de la in- teligencia artificial. El objetivo del mismo es realizar una comparativa de los diferentes frameworks y bibliotecas existentes en el mercado, a día de la realización del presente documento, sobre aprendizaje profundo. Esta comparativa, pretende compararlos us- ando diferentes métricas y una batería heterogénea de pruebas. Adicionalmente se realizará un detallado estudio del arte indicando las técnicas sobre las que sustentan dichas bibliotecas. Se indicarán con ejemplos cómo usar cada una de estas técnicas, las ventajas e inconvenientes de las mismas y sus relaciones. Keywords: aprendizaje profundo, aprendizaje automático frameworks, comparativa v This page has been intentionally left blank. Agradecimientos Probablemente se podría escribir la mitad del documento con agradecimientos a las personas que han influido en que este documento pueda existir. Podría partir de mi abuelo quién me enseñó la informática y lo que se podía lograr con el Amstrad y de ahí hasta el día de hoy, a familia, compañeros, amigos que te influyen, incluso las personas que influyen negativamente, con los que ves que que camino no debes escoger.
    [Show full text]
  • Machine Learning
    1 LANDSCAPE ANALYSIS MACHINE LEARNING Ville Tenhunen, Elisa Cauhé, Andrea Manzi, Diego Scardaci, Enol Fernández del Authors Castillo, Technology Machine learning Last update 17.12.2020 Status Final version This work by EGI Foundation is licensed under a Creative Commons Attribution 4.0 International License This template is based on work, which was released under a Creative Commons 4.0 Attribution License (CC BY 4.0). It is part of the FitSM Standard family for lightweight IT service management, freely available at www.fitsm.eu. 2 DOCUMENT LOG Issue Date Comment Author init() 8.9.2020 Initialization of the document VT The second 16.9.2020 Content to the most of chapters VT, AM, EC version Version for 6.10.2020 Content to the chapter 3, 7 and 8 VT, EC discussions Chapter 2 titles 8.10.2020 Chapter 2 titles and Acumos added VT and bit more to the frameworks Towards first full 16.11.2020 Almost every chapter has edited VT version First full version 17.11.2020 All chapters reviewed and edited VT Addition 19.11.2020 Added Mahout and H2O VT Fixes 22.11.2020 Bunch of fixes based on Diego’s VT comments Fixes based on 24.11.2020 2.4, 3.1.8, 6 VT discussions on previous day Final draft 27.11.2020 Chapters 5 - 7, Executive summary VT, EC Final draft 16.12.2020 Some parts revised based on VT comments of Álvaro López García Final draft 17.12.2020 Structure in the chapter 3.2 VT updated and smoke libraries etc.
    [Show full text]
  • Big Data Management of Hospital Data Using Deep Learning and Block-Chain Technology: a Systematic Review
    EAI Endorsed Transactions on Scalable Information Systems Research Article Big Data Management of Hospital Data using Deep Learning and Block-chain Technology: A Systematic Review Nawaz Ejaz1, Raza Ramzan1, Tooba Maryam1,*, Shazia Saqib1 1 Department of Computer Science, Lahore Garrison University, Lahore, Pakistan Abstract The main recompenses of remote and healthcare are sensor-based medical information gathering and remote access to medical data for real-time advice. The large volume of data coming from sensors requires to be handled by implementing deep learning and machine learning algorithms to improve an intelligent knowledge base for providing suitable solutions as and when needed. Electronic medical records (EMR) are mostly stored in a client-server database and are supported by enabling technologies like Internet of Things (IoT), Sensors, cloud, big data, Deep Learning, etc. It is accessed by several users involved like doctors, hospitals, labs, insurance providers, patients, etc. Therefore, data security from illegal access is crucial especially to manage the integrity of data . In this paper, we describe all the basic concepts involved in management and security of such data and proposed a novel system to securely manage the hospital’s big data using Deep Learning and Block-Chain technology. Keywords: Electronic medical records, big data, Security, Block-chain, Deep learning. Received on 23 December 2020, accepted on 16 March 2021, published on 23 March 2021 Copyright © 2021 Nawaz Ejaz et al., licensed to EAI. This is an open access article distributed under the terms of the Creative Commons Attribution license, which permits unlimited use, distribution and reproduction in any medium so long as the original work is properly cited.
    [Show full text]
  • Outline of Machine Learning
    Outline of machine learning The following outline is provided as an overview of and topical guide to machine learning: Machine learning – subfield of computer science[1] (more particularly soft computing) that evolved from the study of pattern recognition and computational learning theory in artificial intelligence.[1] In 1959, Arthur Samuel defined machine learning as a "Field of study that gives computers the ability to learn without being explicitly programmed".[2] Machine learning explores the study and construction of algorithms that can learn from and make predictions on data.[3] Such algorithms operate by building a model from an example training set of input observations in order to make data-driven predictions or decisions expressed as outputs, rather than following strictly static program instructions. Contents What type of thing is machine learning? Branches of machine learning Subfields of machine learning Cross-disciplinary fields involving machine learning Applications of machine learning Machine learning hardware Machine learning tools Machine learning frameworks Machine learning libraries Machine learning algorithms Machine learning methods Dimensionality reduction Ensemble learning Meta learning Reinforcement learning Supervised learning Unsupervised learning Semi-supervised learning Deep learning Other machine learning methods and problems Machine learning research History of machine learning Machine learning projects Machine learning organizations Machine learning conferences and workshops Machine learning publications
    [Show full text]
  • European Journal of Computer Science and Information
    European Journal of Computer Science and Information Technology Vol.9, No.2, pp.21-28, 2021 Print ISSN: 2054-0957 (Print), Online ISSN: 2054-0965 (Online) COMPARING THE TECHNOLOGIES USED FOR THE APPLICATION OF ARTIFICIAL NEURAL NETWORKS Elda Xhumari PhD Candidate, University of Tirana, Faculty of Natural Sciences, Department of Informatics ABSTRACT: In the past few years, artificial neural networks (ANNs) have become a practice wherever it is necessary to solve problems of forecasting, classification, or control. ANNs are intuitively attractive because they are based on a primitive biological model of nervous systems. In the future, the development of such neurobiological models may lead to the creation of truly thinking computers. The areas of application of neural networks are very diverse - these are text and speech recognition, semantic search, expert systems, and decision support systems, prediction of stock prices, security systems, text analysis, etc. Based on the wide application of artificial neural networks, the application of numerous and diverse technologies can be inferred. In this research paper, a comparative approach towards the above-mentioned technologies will be undertaken in order to identify the optimal technology for each problem area in accordance with their respective advantages and disadvantages. KEYWORDS: artificial neural networks, applications, modules INTRODUCTION Artificial Neural Network (ANN) is a computational method that integrates the concepts of the human biological brain to computers for machine learning. The main purpose is to enable computers to adapt and learn the exact behaviour of human beings. As such, machine learning facilitates the continued processing of information that leads to the capacity to solve complex problems and equations that are beyond human ability (Goncalves et al., 2013).
    [Show full text]
  • Machine Learning Technique Used in Enhancing Teaching Learning Process in Education Sector and Review of Machine Learning Tools Diptiverma #1, Dr
    International Journal of Scientific & Engineering Research Volume 9, Issue 6, June-2018 ISSN 2229-5518 741 Machine Learning Technique used in enhancing Teaching Learning Process in education Sector and review of Machine Learning Tools DiptiVerma #1, Dr. K. Nirmala#2 MCA Department, SavitribaiPhule Pune University DYPIMCA, Akurdi, Pune , INDIA [email protected], [email protected] Abstract: Machine learning is one of the widely rigorous study that teaching learning process has used research area in today’s world is artificial been changing over the time. The traditional intelligence and one of the scope full area is protocol of classroom teaching has been taken to Machine Learning (ML). This is study and out of box level, where e-learning, m-learning, analysis of some machine learning techniques Flipped Classroom, Design Thinking (Case used in education for particularly increasing the Method),Gamification, Online Learning, Audio & performance in teaching learning process, along Video learning, “Real-World” Learning, with review of some machine learning tools. Brainstorming, Classes Outside the Classroom learning, Role Play technique, TED talk sessions Keyword:Machine learning, Teaching Learning, are conducted to provide education beyond text Machine Learning Tools, NLP book and syllabus. I. INTRODUCTION II. MACHINE LEARNING PROCESS: Machine Learning is a science making system Machine Learning is a science of making system (computers) to understand from the past behavior smart. Machine learning we can say has improved of data or from historic data to behave smartly in the lives of human being in term of their successful every situation like human beings do. Without implementation and accuracy in predicting having the same type of situation a human can results.[11] behave or can react to the condition smartly.
    [Show full text]
  • RSNA Mini-Course Image Interpretation Science
    7/31/2017 Can deep learning help in cancer diagnosis and treatment? Lubomir Hadjiiski, PhD The University of Michigan Department of Radiology Outline • Applications of Deep Learning to Medical Imaging • Transfer Learning • Some practical aspects of Deep Learning application to Medical Imaging Neural network (NN) Input Output node node 1 7/31/2017 Convolution neural network (CNN) First hidden Second hidden layer layer Output node Input Image patch filter filter N groups M groups CNN in medical applications First hidden Second hidden layer layer Input Output Image node patch filter filter N groups M groups Chan H-P, Lo S-C B, Helvie MA, Goodsitt MM, Cheng S and Adler D D 1993 Recognition of mammographic microcalcifications with artificial neural network, RSNA Program Book 189(P) 318 Chan H-P, Lo S-C B, Sahiner B, Lam KL, Helvie MA, Computer-aided detection of mammographic microcalcifications: Pattern recognition with an artificial neural network, 1995, Medical Physics Lo S-C B, Chan H-P, Lin J-S, Li H, Freedman MT and Mun SK 1995, Artificial convolution neural network for medical image pattern recognition, Neural Netw. 8 1201-14 Sahiner B, Chan H-P, Petrick N, Wei D, Helvie MA, Adler DD and Goodsitt MM 1996 Classification of mass and normal breast tissue: A convolution neural network classifier with spatial domain and texture images, IEEE Trans.on Medical Imaging 15 Samala RK, Chan H-P, Lu Y, Hadjiiski L, Wei J, Helvie MA, Digital breast tomosynthesis: computer-aided detection of clustered microcalcifications on planar projection images, 2014 Physics in Medicine and Biology Deep Learning Deep-Learning Convolutional Neural Network (DL-CNN) FC Conv1 Conv2 Local1 Local2 Pooling, Norm Pooling, Pooling, Norm Pooling, Input layer Convolution Convolution Convolution Convolution Fully connected Layer Layer Layer (local) Layer (local) layer ROI images Krizhevsky et al, 2012, Adv.
    [Show full text]
  • Convolutional Neural Networks on FPGA and GPU on the Edge: a Comparison
    UPTEC F 20028 Examensarbete 30 hp Juni 2020 Convolutional Neural Networks on FPGA and GPU on the Edge: A Comparison Linus Pettersson Abstract Convolutional Neural Networks on FPGA and GPU on the Edge: A Comparison Linus Pettersson Teknisk- naturvetenskaplig fakultet UTH-enheten When asked to implement a neural network application, the decision concerning what hardware platform to use may not always be easily Besöksadress: made. This thesis studies various relevant platforms with regards to Ångströmlaboratoriet Lägerhyddsvägen 1 performance, power efficiency and usability, with the purpose of Hus 4, Plan 0 providing a basis for such decisions. The hardware platforms which are studied were a GPU, an FPGA and a CPU. The project implements Postadress: Convolutional Neural Networks (CNN) on the different hardware Box 536 751 21 Uppsala platforms using several tools and frameworks. The final implementation uses BNN-PYNQ for the implementation on the FPGA and Telefon: CPU, which provided ready-to-run code and overlays for quantized CNNs 018 – 471 30 03 and fully connected neural networks. Next, these networks are copied Telefax: using TensorFlow, and optimized to FP32, FP16 and INT8 precision 018 – 471 30 00 using TensorRT for use on the GPU. The results indicate that the FPGA outperforms the GPU with a factor of 100 for the CNN networks, and a Hemsida: factor of 1000 on the fully connected networks with regards to http://www.teknat.uu.se/student inference speed. For power efficiency, the FPGA again outperforms the GPU. The thesis concludes that for a neural network application, an FPGA is preferred if performance is a priority.
    [Show full text]
  • Design and Implementation of a Domain Specific Language for Deep Learning Xiao Bing Huang University of Wisconsin-Milwaukee
    University of Wisconsin Milwaukee UWM Digital Commons Theses and Dissertations May 2018 Design and Implementation of a Domain Specific Language for Deep Learning Xiao Bing Huang University of Wisconsin-Milwaukee Follow this and additional works at: https://dc.uwm.edu/etd Part of the Artificial Intelligence and Robotics Commons, and the Electrical and Computer Engineering Commons Recommended Citation Huang, Xiao Bing, "Design and Implementation of a Domain Specific Language for Deep Learning" (2018). Theses and Dissertations. 1829. https://dc.uwm.edu/etd/1829 This Dissertation is brought to you for free and open access by UWM Digital Commons. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of UWM Digital Commons. For more information, please contact [email protected]. DESIGN AND IMPLEMENTATION OF A DOMAIN SPECIFIC LANGUAGE FOR DEEP LEARNING by Xiao Bing Huang A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Engineering at The University of Wisconsin-Milwaukee May 2018 ABSTRACT DESIGN AND IMPLEMENTATION OF A DOMAIN SPECIFIC LANGUAGE FOR DEEP LEARNING by Xiao Bing Huang The University of Wisconsin-Milwaukee, 2018 Under the Supervision of Professor Tian Zhao Deep Learning (DL) has found great success in well-diversified areas such as machine vision, speech recognition, big data analysis, and multimedia understanding recently. However, the existing state-of-the-art DL frameworks, e.g. Caffe2, Theano, TensorFlow, MxNet, Torch7, and CNTK, are programming libraries with fixed user interfaces, internal represen- tations, and execution environments. Modifying the code of DL layers or data structure is very challenging without in-depth understanding of the underlying implementation.
    [Show full text]
  • Machine Learning Singularity
    Machine Learning Singularity For the journal, see Machine Learning (journal). "Computing Machinery and Intelligence" that the ques- tion “Can machines think?" be replaced with the ques- [1] tion “Can machines do what we (as thinking entities) can Machine learning is a subfield of computer science [10] that evolved from the study of pattern recognition and do?" computational learning theory in artificial intelligence.[1] In 1959, Arthur Samuel defined machine learning as a “Field of study that gives computers the ability to learn 1.1 Types of problems and tasks without being explicitly programmed”.[2] Machine learn- ing explores the study and construction of algorithms that Machine learning tasks are typically classified into three can learn from and make predictions on data.[3] Such al- broad categories, depending on the nature of the learn- gorithms operate by building a model from an example ing “signal” or “feedback” available to a learning system. training set of input observations in order to make data- These are[11] driven predictions or decisions expressed as outputs,[4]:2 rather than following strictly static program instructions. • Supervised learning: The computer is presented Machine learning is closely related to (and often overlaps with example inputs and their desired outputs, given with) computational statistics; a discipline which also fo- by a “teacher”, and the goal is to learn a general rule cuses in prediction-making through the use of comput- that maps inputs to outputs. ers. It has strong ties to mathematical optimization, which delivers methods, theory and application domains to the • Unsupervised learning: No labels are given to the field.
    [Show full text]
  • Review on Blue Brain for Novice Kowshalya .G
    ISSN XXXX XXXX © 2017 IJESC Research Article Volume 7 Issue No.8 Review on Blue Brain for Novice Kowshalya .G. MCA Assistant Professor Department of Computer Science Sri GVG Visalakshi College for Women, Udumalpet, India Abstract: Uploading the human brain into the supercomputer is called Blue Brain. So the machine can function as human and it can take decisions. The Cognitive learning method is used for simulation here. Based on the working principle of human brain, virtual brain is modeled. Blue brain will not come under the category of Artificial intelligence (AI),it comes under the subset of AI that is deep machine learning. The more processing power is needed to here to simulate the whole brain. With advantages there also demerits associated with this technology. The major objective of the Blue Brain Project is to crack open secrets of how the brain rewires itself every moment of its existence. The resultant knowledge would lead to a new breed of super computers. Keywords: Cognitive Learning, Machine Learning, Nano robots, Pattern Recognition, Liquid Computing. I. INTRODUCTION 2.2 Functioning of human brain: Human brain is the most valuable creation of God. Because of 2.2.1 Sensory Input: death, our body and brain get destroyed. It is possible to transplant our body organs and make them alive even after our The getting of information from our surroundings is called death. It is also possible to make our brain (i.e. intelligence) to sensory input. For example, if we smell a rose or our eyes see alive even after death. Human brain gets copied into computer, something, suppose our hands touch a hot water, then these so the intelligence of anyone can be loaded into the computer.
    [Show full text]
  • Diving Deep Into Deep Learning: History, Evolution, Types and Applications
    International Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-9 Issue-3, January 2020 Diving Deep into Deep Learning: History, Evolution, Types and Applications Deekshith Shetty, Harshavardhan C A, M Jayanth Varma, Shrishail Navi, Mohammed Riyaz Ahmed Abstract: Although Machine Learning (ML) has become synonymous for Artificial Intelligence (AI); recently, Deep Learning (DL) is being used in place of machine learning persistently. If statistics is grammar and machine learning is poetry then deep learning is the creation of Socrates. While machine learning is busy in supervised and unsupervised methods, deep learning continues its motivation for replicating the human nervous system by incorporating advanced types of Neural Networks (NN). Due to its practicability, deep learning is finding its applications in various AI solutions such as computer vision, natural language processing, intelligent video analytics, analyzing hyperspectral imagery from satellites and so on. Here we have made an attempt to demonstrate strong learning ability and better usage of the dataset for feature extraction by deep learning. This paper provides an introductory tutorial to the Fig. 1. Google trend search result (past 15 years) for AI, domain of deep learning with its history, evolution, and th introduction to some of the sophisticated neural networks such as ML, DL as on 14 March 2019. Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). This work will serve as an introduction to the order to mimic neural behavior many attempts have been amazing field of deep learning and its potential use in dealing made and models have been created.The first among such with today’s large chunk of unstructured data, that it could take models was the single layered perceptron [5].
    [Show full text]