No student devices needed. Know more
85 questions
Which can help to prevent exploding gradients?
Standardization
Batch normalization
Lowering the learning rate
Standardization
___________ can help keep ReLU units from dying.
Backpropagation
Lowering the learning rate
Batch normalization
Standardization
Updating the weights
Check the standard components of neural network.
A set of nodes, analogous to neurons, organized in layers.
A set of weights representing the connections between each neural network layer and the layer beneath it. The layer beneath may be another neural network layer, or some other kind of layer.
A set of biases, one for each node.
An activation function that transforms the output of each node in a layer. Different layers may have different activation functions.
“Convolutional Neural Networks can perform various types of transformation (rotations or scaling) in an input”. Is the statement correct True or False?
True
False
In custom modelling, which cases are true?
Easy to get started
Requires ML expertise
Data is accessible to provider
Robust support, auto fine-tuning
Expensive procedure
Compute the following from the confusion matrix2. Sensitivity
85.91
87.51
95.23
85.79
91.78
In Automated ML, which cases are true?
Robust Enterprise support
Full control over parameter tuning
Unlimited use-cases
Difficult to extend
Complete customization facility
Which of the following option is true?
Linear Regression errors values has to be normally distributed but in case of Logistic Regression it is not the case
Logistic Regression errors values has to be normally distributed but in case of Linear Regression it is not the case
Both Linear Regression and Logistic Regression error values have to be normally distributed
Both Linear Regression and Logistic Regression error values have not to be normally distributed
Standardisation / Normalization of features is required before training a Logistic Regression.
True
False
An enormous data set with randomly sampled examples probably :
does not contain redundant data
tends not to carry much more predictive value than large batches
has more predictive value than a medium size data set
What happens if you enter a negative value for a feature cross?
The model will not separate the classes
Predictions will be completely wrong
Both
None
An expensive robotic chicken crosses a very busy road a thousand times per day. An ML model evaluates traffic patterns and predicts when this chicken can safely cross the street with an accuracy of 99.99%.
Here accuracy value suggest that the ML model is doing a good job.
True
False
Consider a classification model that separates email into two categories: "spam" or "not spam." If you raise the classification threshold, what will happen to precision?
Probably increase.
Definitely increase
Probably decrease.
Definitely decrease.
Which of the following hyper parameter(s), when increased may cause random forest to over fit the data?
1. Number of Trees2. Depth of Tree3. Learning Rate
Only 1
Only 2
Only 3
Both 1 & 2
Both 1 & 3
Which of the following is an example of a deterministic algorithm?
PCA
K-Means
Both
None
Minimizing loss+complexity is called empirical risk minimization
True
False
In practice, machine learning models frequently cross continuous features.
True
False
Imagine a linear model with 100 input features:
* 10 are highly informative.
* 90 are non-informative.Assume that all features have values between -1 and 1.
Which of the following statements are true?
L2 regularization will encourage many of the non-informative weights to be nearly (but not exactly) 0.0.
L2 regularization will encourage most of the non-informative weights to be exactly 0.0.
L2 regularization may cause the model to learn a moderate weight for some non-informative features.
Different cities in California have markedly different housing prices. Suppose you must create a model to predict housing prices. Which of the following sets of features or feature crosses could learn city-specific relationships between roomsPerPerson and housing price?
One feature cross: [binned latitude X binned longitude X binned roomsPerPerson]
Two feature crosses: [binned latitude X binned roomsPerPerson] and [binned longitude X binned roomsPerPerson]
One feature cross: [latitude X longitude X roomsPerPerson]
Three separate binned features: [binned latitude], [binned longitude], [binned roomsPerPerson]
Check all the good features.
house_type: victorian
unique_house_id: 8SK982ZZ1242Z
house_age_years: 27
house_age: 851472000
city_id: "br/sao_paulo"
Which method is good for handling extreme outliers?
Logarithmic scaling
Binning
Winsorizing
Clipping
A Pearson correlation between two variables is zero but, still their values can still be related to each other.
True
False
Models cannot multiply strings by the learned weights, that's why we can not use feature engineering to convert strings to numeric values.
True
False
If we incorporate index numbers (i.e. 1, 2, 3, 4) instead of string values or categorical data directly into our model, it won't impose any constraints that might be problematic.
True
False
In case of pneumonia, how you can annotate?
On a numerical scale
0-n for their confidence that an image contains pneumonia symptoms or not.
String value like Pneumonia / Not-pneumonia
By color i.e. red for Pneumonia and Green for Not-pneumonia
What you should include in case of uncertainty in an annotation?
Serious case
Unknown
Healthy case
Other
State whether the statements are True or False.
Statement A: When the hypothesis space is richer, overfitting is more likely.
Statement B: When the feature space is larger, overfitting is more likely.
False, False
True, False
True, True
False, True
You have generated data from a 3-degree polynomial with some noise. What do you expect of the model that was trained on this data using a 5-degree polynomial as function class?
Low bias, high variance
High bias, low variance
Low bias, low variance
High bias, low variance
What will happen when eigenvalues are roughly equal in PCA?
PCA will perform outstandingly
PCA will perform badly
Can’t Say
None of above
PCA works better if there is?
(i) A linear structure in the data
(ii) If the data lies on a curved surface and not on a flat surface
(iii) If variables are scaled in the same unit
1 and 2
2 and 3
1 and 3
1 ,2 and 3
Select the option(s) that complete the following sentence correctly. Tf.keras is _______
the TensorFlow variant of the open-source Keras API
an implementation of the Keras API
the Keras variant of the open-source TensorFlow API
an end-to-end open-source platform for machine learning
a high-level API for TensorFlow
If you want to create a numpy array using random floating point number between a range, what you will use?
np.arrange(lower_bound_floating_num. higher_bound_floating_num)
np.random.uniform(ower_bound_floating_num. higher_bound_floating_num)
np.random.uniform(ower_bound_floating_num. higher_bound_floating_num, array_shape)
None
If you want to find the number of rows and columns of dataset what you will use?
dataset.size()
dataset.shape
dataset.shape()
dataset.size
How to populate an array with a sequence of numbers in numpy?
np.arrange(5, 12)
np.arange(5, 12)
np.random.randint(5, 12)
np.random.random(5, 12)
NumPy uses a trick called _______ to virtually expand the smaller operand to dimensions compatible for linear algebra.
list comprehension
broadcasting
matrix multiplication
lambda function operation
Underfitting occurs due to ______
High bias
High variance
Low bias
Low variance
As a product manager, problems can come from ____
Pressure from investor
Upset users or customers
Lost revenue
Grumpy Engineers
Struggling sales and customer teams
When we talk about impact, we often want to think about the primary users.
True
False
What are the steps to solve unwanted bias?
Awareness
Resource engagement
Data management
Iteration and learning
Unbalanced selection of sorts data generate _____
Selection bias
Data bias
Model bias
Annotation bias
Which method we should use in case of annotating very very small object in video annotating task?
Computer vision
Image detection algorithm
Linear interpolation
Convolutional neural network
In supervised learning, a machine learning algorithm builds a model by examining many examples and attempting to find a model that minimizes loss; this process is called _______
activation function
cost function
empirical risk minimization
gradient descent
lagloss function
When performing gradient descent on a large data set, which of the following batch sizes will likely be more efficient?
Batch gradient descent
Stochastic gradient descent
Mini-batch stochastic gradient descent
Suppose an online shoe store wants to create a supervised ML model that will provide personalized shoe recommendations to users. That is, the model will recommend certain pairs of shoes to Marty and different pairs of shoes to Janet. The system will use past user behavior data to generate training data. Which of the following statements are true?
"Shoes that a user adores" is a useful label.
"The user clicked on the shoe's description" is a useful label.
"Shoe size" is a useful feature.
"Shoe beauty" is a useful feature.
Best practices of A/B testing are.............
Deciding on performance
Deciding on test type based on your performance metric
Choosing a maximum effect size you want to detect
Determining the sample size
Running the test until sample size is reached
Relu helps to make model linear.
True
False
Convolutional layer helps to reduce dimension of images.
True
False
Pooling layer helps to reduce the dimension of input.
True
False
Dense layer is used for classification in CNN.
True
False
Stride in CNN mean the size of pixel shift over the input matrix.
True
False
Checkout all the annotation tools for object detection.
Appen
LabelIMG
Prodigy
Datatrunks
Hive
Check all the feature extractors:
Unet
Vgg16
Resnet
inception
yolo
Which is the best for realtime object detection in case of speed?
Yolo v4
Yolo v3
Faster RCNN
SSD
RCFN
Which method is fast?
Stemming
Lemmatization
Word2Vec
What are the problems of RNN?
Vanishing Gradient Problem
Weight Initialization Problem
Problem in Backpropagation due to activation function
Exploding Gradient Problem
What is the solution of the problems of RNN?
Bidirectional RNN
LSTM
Word Embedding
Stacked RNN
Why dropout is used?
To prevent underfitting
To prevent overfitting
To prevent exploding gradient descent
To prevent vanishing gradient descent problem
Suppose there is an image that contains humans. If you actually paint the pixels corresponding to humans in this image, then this is called semantic segmentation.
True
False
SSD performs well in large objects.
True
False
YOLO performs well on small objects.
True
False
Which is the region based detector?
RCNN
YOLO
SSD
SSD uses fixed grid regression.
True
False
Stemming can be used in _______
Sentiment Classification
Chatbots
Spam Classifier
Q/A App
What are the better alternative solution of Bag Of Words?
Lemmatization
TF-IDF
Word to Vec
What are the disadvantages of TF-IDF?
Semantic information is not stored
Gives importance to uncommon workds
Chance of overfitting
Equal weight edge
VGG16 is ______
Feature extractor
RNN model
LSTM model
Computer Vision algorithm
RCNN has 3 models.
True
False
UNET is used for ________
Semantic segmentation of image
Image classification
Object detection
___________ can help keep ReLU units from dying.
Backpropagation
Lowering the learning rate
Batch normalization
Standardization
Updating the weights
When the number of classes is large enough, we can use _______
candidate sampling
full softmax
If we want to safe from underfitting, then regularization is the only way.
True
False
The cost function of linear regression can be used in logistic regression.
True
False
K-means clustering is semi-supervised machine learning system.
True
False
Which can be used as dimensionality reduction algorithm in unsupervised learning?
SMOTE
PCA
LDA
Pooling layers
Which is used to get optimal value in k-means clustering?
PCA
SMOTE
RMSE
Confusion Matrix
Elbow Method
Which will be better option for imbalanced dataset if the dataset is not huge?
Undersampling
Oversampling
Which section backpropagation updates in CNN?
Maxpooling layer
Padding
Feature extraction layer
Kernel
Which model or algorithms does not need scaling?
CNN
Logistic Regression
Reinforcement learning
Xgboost
To pick up the high intensity from the feature map, which should be used in CNN?
Padding
Conv layer
Pooling
None of them
Targetted marketing, Recommended Systems, and Customer Segmentation are applications in ...
Unsupervised Learning: Clustering
Supervised Learning: Classification
Reinforcement Learning
Unsupervised Learning: Regression
Fraud Detection, Image Classification, Diagnostic, and Customer Retention are applications in ...
Unsupervised Learning: Clustering
Supervised Learning: Classification
Reinforcement Learning
Unsupervised Learning: Regression
This picture shows a result of ...
Supervised Learning: Classification
Unsupervised Learning: Regression
Unsupervised Learning: Prediction
Supervised Learning: Regression
This picture shows an application of ...
Supervised Learning: Classification
Unsupervised Learning: Clustering
Unsupervised Learning: Prediction
Supervised Learning: Regression
Machine Learning has various function representation, which of the following is not function of symbolic?
Decision Trees
Rules in propotional Logic
Hidden-Markov Models (HMM)
Rules in first-order predicate logic
Machine Learning has various search/ optimization algorithms, which of the following is not evolutionary computation?
Perceptron
Genetic Algorithm (GA)
Neuro Evolution
Genetic Programming (GP)
Explore all questions with a free account