Data Science

The post is about a Deep Learning Quiz. There are 20 multiple-choice questions covering topics related to the pros and cons of Deep Learning (DL), DL and Machine Learning (ML) Methods for the improvement of output variables, structured data, TensorFlow, AlphaGo, Microsoft CNTK, Google’s TPU, etc. Let us start with the Deep Learning Quiz now.

Online Deep Learning Quiz with Answers

1. Among the following statements on Google’s TensorFlow, which is incorrect?

 The initial release of TensorFlow was v0.5.0 that was released in November of 2015

 TensorFlow does not currently include support for HDFS (Hadoop Distributed File System) based Hadoop big data

 TensorFlow is the most popular ML (Machine Learning) language in GitHub as of June 2017

 A “tensor” is an organized multidimensional array of data values used in physics and engineering in modeling

2. Which of the following is not a DL (Deep Learning) open source software?

 Microsoft CNTK (Cognitive Toolkit)

 Caffe

 TensorFlow

 MATLAB

 Keras

 Theano

3. Among the following statements on the pros and cons of DL (Deep Learning), which is incorrect?

 One of the cons of DL systems is the expensive cost

 One of the pros of DL systems is the short training time

 One of the pros of DL systems is the reduced need for feature engineering and that DL systems are adaptable to multiple problem domains

 One of the cons of DL based decision assisting systems is that a DL system could result in racial discrimination and other social issues, so extra caution in decision making is needed

4. Among the following reasons on why DL (Deep Learning) is popular now, which is incorrect?

 As deep learning is a software innovation, it has the big advantage of being independent of hardware improvements

 Easy-to-use open source software with advanced library models and pre-trained open models are a driving source of DL systems

 Greater availability of higher quality of data through intelligent data filters and databases is one of the reasons DL systems are becoming more accurate and reliable

 Declining computational costs in cloud computing, which include MEC (Mobile Edge Computing) and Fog computing is one of the reasons DL systems are showing much improvement

5. Among the following descriptions on Google’s AlphaGo, which is incorrect?

 AlphaGo Lee used 50 first generation TPUs (Tensor Processing Units)

 AlphaGo Fan used 1,202 CPUs (Central Processing Units) and 176 GPU (Graphics Processing Units)

 AlphaGo Master used a second generation TPU that has a processing capability of 11.5 PFLOPS

 IBM’s Deep Blue was the first AI (Artificial Intelligent) system to defeat professional human Go players, and AlphaGo was the second

6. Among the following methods of how DL (Deep Learning) and ML (Machine Learning) can help to improve sales growth, which is incorrect?

 For new business strategies, use ML to generate synthetic data through extrapolation and projection of existing data

 Develop ML based optimized propensity strategies to enhance up-sell, cross-sell, and omnichannel selling

 Use ML to help analyze the structured, semi-structured, and unstructured data from the IoT and smart device network

 When a miscalculation is detected from the ML system, do not use the ML system in the correction process to avoid additional system mistakes

7. Among the following descriptions on Google’s TPU (Tensor Processing Unit), which is incorrect?

 TPU is Google’s ML (Machine Learning) ASIC (Application-Specific Integrated Circuit) processor designed specifically for TensorFlow processing

 First generation TPUs have a total processing capability of 700 GFLOPS

 Second generation TPUs have a total processing capability of 11.5 PFLOPS

 Second generation TPUs consists of 256 chip pods (64 integrated 4-chip modules) consisting of each 45 TFLOPS computing modules

8. Among the following descriptions of the Object Localization contest (top-5) winners of the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?

 The 2012 winner AlexNet used a DL CNN (Convolution Neural Network) to achieve a 83.6% performance, which was a significant improvement compared to 2011

 The 2015 winner was Microsoft ResNet with a 96.5% accuracy, which exceeded the 94.9% human accuracy level for the first time

 The 2014 winner was GoogleNet (Inception-v1) with a 93.3% accuracy performance, which exceeded the 90% accuracy level for the first time

 The 2011 winner Xerox was the first to use DL (Deep Learning) technology in achieving a remarkable 74.2% accuracy performance

9. Among the following application areas corresponding to data types that are used in businesses with DL (Deep Learning) and ML (Machine Learning) technology, which is incorrect?

 Theme detection DL & ML technology is used in finance estimation and prediction systems

 Sentiment analysis DL & ML technology is used in CRM (Customer Relationship Management) of customers and potential customers

 Due to the training time required, threat detection systems using DL & ML technology used in airport security systems are processed in non-real-time database systems

 Motion detection DL & ML technology is used in games and robotics

10. Among the following statements on Microsoft CNTK (Cognitive Toolkit), which is incorrect?

 In CNTK, program development can be conducted with Python or C++

 BrainScript is used in defining custom networks (improved configuration scripts)

 CNTK provides optimized runtime for DL (Deep Learning) neural network training and testing

 Microsoft originally introduced the CNTK (Computational Network Toolkit) in 2016

11. Among the following descriptions on NVIDIA DGX-1, which is incorrect?

 DGX-1 supports interfaces with clouds, data centers, workstations, and embedded platforms

 NVIDIA Deep Learning SDK (Software Development Kit) can be used on a DGX-1 system

 DGX-1 has a processing capability of 170 GFLOPS

 DGX-1’s DL (Deep Learning) software includes a precision DNN (Deep Neural Network) stack

12. Among the following statements on Google’s TensorFlow, which is incorrect?

 TensorFlow supports CUDA, which was developed by Nvidia and stands for “Compute Unified Device Architecture”

 TensorFlow includes support of Java APIs (Application Programming Interfaces) for Windows

 TensorFlow currently supports Python and Microsoft Windows, but iOS is not supported yet

 TensorFlow supports cuDNN which stands for “CUDA Deep Neural Network library,” which is used for DL (Deep Learning) neural network designing on Nvidia GPUs (Graphics Processing Units)

13. Among the following descriptions on Google’s AlphaGo, which is incorrect?

 AlphaGo Lee used 50 first generation TPUs (Tensor Processing Units)

 AlphaGo Fan used 1,202 CPUs (Central Processing Units) and 176 GPU (Graphics Processing Units)

 AlphaGo Master used 256 second generation TPUs

 AlphaGo is the first Go program to defeat professional human Go players

14. Among the following mappings of CNTK Network Graph Nodes, which is incorrect?

 CNTK Output: OutputNode

 Training Labels: LabelNode

 Result Evaluation: EvalNode

 Training: CriterionNode

 CNTK Input: InputNode

15. Among the following statements on Google’s TensorFlow, which is incorrect?

 TensorFlow includes support for HDFS (Hadoop Distributed File System) big data systems

 TensorFlow currently supports Python, iOS, and Microsoft Windows OS (Operating Systems)

 The initial release of TensorFlow was v0.1 in 2005

 A “tensor” is an organized multidimensional array of data values used in physics and engineering in modeling

16. Among the following descriptions on Google’s AlphaGo, which is incorrect?

 Training of AlphaGo Master was based on supervised learning only

 AlphaGo’s Value Network predicts the Go game winner

 AlphaGo combines ATS (Advanced Tree Search) and DNNs (Deep Neural Networks) technology

 AlphaGo’s Policy Network selects the next move to play in the Go game

17. Based on the lecture “Characteristics of Businesses with DL & ML,” among the following data types used in businesses with DL (Deep Learning) and ML (Machine Learning) technology, which is not a structured data type?

 Time Series (i.e., discrete data values indexed in time order)

 RDBMS (Relational Database Management System)

 Sound & text

 RDSMS (Relational Data Stream Management System)

18. Among the following descriptions on the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?

 Object Localization contest (which is also known as the “top-5”) evaluates contestants with a test image set of 150,000 photographs in listing the 5 highest confident labels of each test image

 ILSVRC is an annual contest that started in 2010 for object category classification and detection

 The original competition of ILSVRC was the video Object Detection competition

 The Object Detection contest on “Images” determines the contestant program with the highest accuracy in annotated class labels, confidence scores, and bounding boxes based on identification of 200 basic-level categories

19. Among the following descriptions of the Object Localization contest (top-5) winners of the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?

 In 2012, AlexNet won the top-5 contest with a 83.6% accuracy performance using a 8 layer DL CNN (Convolution Neural Network)

 In 2015, Microsoft ResNet won the top-5 contest with a 96.5% accuracy performance using a 152 layer DL neural network

 Because in 2015, the top-5 contest was won with a 96.5% accuracy, which exceeded the 94.9% human accuracy level, the top-5 contest was replaced with the Object Detection challenges on Images and Videos from 2016

 In 2014, GoogleNet (Inception-v1) won the top-5 contest with a 93.3% accuracy performance using a 22 layer DL neural network

 In 2011, Xerox won the top-5 contest with a 74.2% accuracy performance

20. Based on the lecture “What is Deep Learning & Machine Learning,” among the following computer performance units described, which is incorrect?

 Some high-performance PCs (Personal Computers) can perform at 100 GFLOPS levels

 FLOPS is a floating-point computation based performance measure, that stands for “FLoating-Point Operations Per Second”

 Humans perform at an average of 100 FLOPS

 IPS is a number of operations based computation performance measure, that stands for “Instructions Per Second”

 Loading …

Online Deep Learning Quiz with Answers

• Among the following reasons on why DL (Deep Learning) is popular now, which is incorrect?
• Among the following statements on the pros and cons of DL (Deep Learning), which is incorrect?
• Among the following methods of how DL (Deep Learning) and ML (Machine Learning) can help to improve sales growth, which is incorrect?
• Based on the lecture “Characteristics of Businesses with DL & ML,” among the following data types used in businesses with DL (Deep Learning) and ML (Machine Learning) technology, which is not a structured data type?
• Among the following application areas corresponding to data types that are used in businesses with DL (Deep Learning) and ML (Machine Learning) technology, which is incorrect?
• Which of the following is not a DL (Deep Learning) open source software?
• Among the following statements on Google’s TensorFlow, which is incorrect?
• Among the following descriptions on Google’s AlphaGo, which is incorrect?
• Among the following descriptions of the Object Localization contest (top-5) winners of the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?
• Among the following statements on Google’s TensorFlow, which is incorrect?
• Among the following statements on Google’s TensorFlow, which is incorrect?
• Among the following statements on Microsoft CNTK (Cognitive Toolkit), which is incorrect?
• Among the following mappings of CNTK Network Graph Nodes, which is incorrect?
• Among the following descriptions on NVIDIA DGX-1, which is incorrect?
• Among the following descriptions on Google’s AlphaGo, which is incorrect?
• Among the following descriptions on Google’s AlphaGo, which is incorrect?
• Among the following descriptions on Google’s TPU (Tensor Processing Unit), which is incorrect?
• Among the following descriptions on the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?
• Among the following descriptions of the Object Localization contest (top-5) winners of the ILSVRC (ImageNet Large Scale Visual Recognition Challenge), which is incorrect?
• Based on the lecture “What is Deep Learning & Machine Learning,” among the following computer performance units described, which is incorrect?

Take a Test about Exploratory Data Analysis Quiz