Huawei AI Exam
the limitation of a single-layer perceptron is that it cannot solve the XOR problem, feed forward neural network can be represented by a directed acyclic graph, there is no connection between neurons in the same layer of the feedforward neural network.
Common perception about neural networks
sigmoid, tanh, and relu
Commonly used activation functions.
Dark blue
Computer that defeated Kasparov, the world chess king in May 1997 with a score of 3.5 and 2.5
Images
Convolutional neural networks are more suitable in what kind of problems?
Tensor
Data type used in Tensorflow
DVPP, AI Calculation Engine, Chip System control CPU
Deflated AI The logical architecture of the process includes what options?
forget, input, and output door
LSTM network has the following architecture
one
Linear relationship between exponential object are expressed in what form?
Grid search, random search and model-based hyper parameter optimization
What are the methods used in automatic hyperparameter optimization algorithm?
Application enable, Open source framework and chip enable
What are the options to the scope of Huawei's Full-stack solution?
Speed up algorithm convergence, reduce the difficulty of manual parameter setting, avoid local extremes
What are the purpose of an optimizer?
Artificial Intelligence, Web Development and Game Development
What are the scenarios suitable for Python
explicitly call the session to generate function, explicitly call session to close function, and through the Python context manager.
What are the session modes used by Tensorflow?
pruning, feature selection, decision tree generation
What are the steps in decision tree construction process?
new ways of collaboration, new programming paradigm, new execution mode
What are the technological innovations of mindspore?
GMM, clustering, and association rules
What are unsupervised learning algorithms?
What company launched Pytorch first?
Huawei HiAI Engine
What engine can combine multiple AI Ability and App integration?
loss function
What function of logistic regression is the cross-entropy loss function
Historical data
What input in ML model needs to train itself and predict the unknown?
col of matrix A = row of matrix B
What is the condition in multiplying matrices A and B?
placeholder
What mechanism is used by Tensorflow to avoid excessive input data?
Gradient descent
What methods is used to minimize the loss function?
tf.nn, tf.layers, tf.contrib
What modules are in the tensorflow library?
know the actual problem, mathematical model, numerical calculation method, programming, and computer calculation results.
What processes are involved in solving actual problems using a computer?
accuracy and loss
model:model.compile(optimer='Adam',loss="categorical.crossentropy', metrics=[tf.keras.metrics.accuracy]), it will output
Machine Learning Service
A data mining and analytics platform service that helps quickly discover data patterns and build predictive models through machine learning techniques and deploy them as predictive analytics solutions
While loop, for loop, and nested loop statement
Applications for the break statement in Python
28 x 28
Enter a 32 x 32 with size 5x5 step size 1.
Logistic regression, support vector machine, decision tree
Examples of Supervised Learning
more than one
Face comparison service need how many images
Big data, computing ability, and algorithm theory
Factors that promote development of AI?
TRUE
Features are interpretable in traditional machine learning but weak in deep learning
GBDT Algo is easier to underfit than random forest, Random forest is calculated in parallel unlike GBDT, GBDT and random forest are built on CART based on the tree.
GBDT vs Random Forest Algo
type for text and table time represents the form recognition area.
General form identification service returned types are
tf.keras.datasets
How to view the keras built in dataset?
crowd statistics program, policy query scheme based on knowledge graph, vehicle identification scheme and intrusion recognition scheme
Huawei Cloud EI Intelligent platform it provides solutions to
rich library of ML algorithms, ml program is intuitive and easy to use and uses distributed and scalable big data computing engine.
Huawei Machine Learning Service is a one-stop platform that supports
HiAI Foundation
Huawei's chip support HUAWEI HiAI came from what module?
underfit
If a model has a large deviation on the test set and small variance, the model is
base 64 coded
Image data of the passport recognition should be in what base code?
/v1.0/image/tagging
Image label service URI?
Regular term
It can be added to logistic regression to avoid overfitting.
Density function
It can only describe the value rule of continuous random variables.
Distribution law
It can only describe the value rule of discrete random variables.
Distribution function
It describes the value rule of random variables.
continue
It is used to jump out of the loop and then continue to execute the next round of statements.
Automatic parallel
MindSpore has handled this challenge in achieving efficient distributed training for ultra-large scale models.
independent and identical distribution
Naive Bayes requires what kind of sample features?
Fully connected and recurrent neural networks
Neural networks that share weights
Probability theory
Not central or linear algebra
Function, Module, Number, Character String
Options belongs to the Python object
Alpha Go, Self-driving, Voice input and Huawei Mate mobile phone
Products related to AI
Support dynamic graph, automatic derivative and GPU accelerate
PyTorch has the following functions
numbers, string, list, tuple, dictionary
Python data types
Python module
Python file that ends with .py and contains python objects, definitions and Python statements. It also allows to logically organize Python code snippets, and assign relevant code to a module can make your code better and easier to understand.
Sequence data
RNN used what kind of data?
Building calculation chart, input tensor and generate sessions
Tensorflow build process are
True
Tensorflow supports multiple programming languages
Approaches it
The ML model cannot represent true data distribution on a theoretical level but it only
machine learning, deep learning, graph engine and batch processing
The basic platform services of huawei cloud e1 enterprise intelligence are
DistBelief
The basis of Tensorflow as a second generation of AI learning system by Google.
Mean Square Loss Function
The commonly used loss function in deep learning is
The company who developed Tensorflow.
userfriendly, modular combination, and easy to expand
The main advantages of tensorflow 2.0 middle Keras
import
The only common in Python 2 and 3
Linear correlation
The ration of change is constant
Artificial Intelligence
The research and development of theories, methods and application systems used to simulate, extend and expand human intelligence of a new technological science.
Machine Learning
The research field in artificial intelligence which mainly studies how computers automatically aqcuire knowledge
result
The result of the voice recognition service is stored in which field?
Parameter
The use gradient descent method to continuously update the value which makes the loss function minimization.
Relu function
To handle gradient vanishing, we can alleviate by choosing what function?
7x24
Traffic agent can achieve hour full time/global traffic awareness?
import time, from time import *, import time as t
Types of module import in Python
Huawei Ascend AI Chip
What AI chip is typical representative of a neural network processor
Sigmoid and Tanh functions
What activation functions are prone to vanishing gradient problems?
Dynamic Information in serialized data
What are captured by Recurrent neural networks?
RNN, LSTM, GRU
What are examples of cylic neural networks?
Knowledge map, machine translation and semantic understanding
What are specific technology in AI?
Multidimensional arrays with a uniform dtype
What are stored in a tensor?
Action, Perception and cognition
What are the categories of AI?
Legibility, Simplicity, and Scalability
What are the characteristics of a Python code?
Explanatory, Object-oriented and Dynamic data type
What are the characteristics of the Python language?
Adam, Adagrad, SGD, Momentum
What are the common deep learning model training processors?
Gaussian, Polynomial, Sigmoid and Linear Kernel Functions
What are the commonly used kernel functions in SVM?
Each test was repeated under the same conditions, there are only two possible outcomes for each trial, the results of each trial are independent of each other.
What are the conditions in a n-fold bernoulli trial?
Convolutional layer, pooling layer, fully connected layer
What are the convolutional neural network structures?
Natural Language processing, computer vision and speech recognition
What are the current technology directions of artificial intelligence
Traditional rule-based methods, rules can be explicitly clarified manually and explicitly programming to solve problems. On ML, mapping relationship is automatically learnied.
What are the difference between machine learning algorithms and traditional rule-based methods?
Business applications, Training, and Reasoning
What are the divisions of AI Chips?
Searched face similarity, searched faces id, searched face position
What are the item results returned when using face search service?
fully connected layer, convolutional layer, and pooling layer
Which layers play the role of dimensionality reduction?
Random number, constants and calculated from initial values of other variables.
Which methods are supported in Tensorflow to define variables?
machine translation, speech recognition, and text generation
Which scenarios are included in a recurrent neural network?
1956
Year when the AI was first introduced
Python function
created using the def keyword followed by the function name and parentheses, parameters are placed in parentheses, function content starts with a colon need to be indented. return the results with return and function ends.
tensors
in tensorflow data are represented in?
Activation function
it is partly a nonlinear function, partly a linear function
math, nn and array
what are MindSpore common operations?
comprehensive interdisciplinary and marginal disciplines
what is the subject of artificial intelligence?
