Introduction to Deep Learning

Karteikarten und Zusammenfassungen für Introduction to Deep Learning an der TU München

Arrow Arrow

Komplett kostenfrei

studysmarter schule studium
d

4.5 /5

studysmarter schule studium
d

4.8 /5

studysmarter schule studium
d

4.5 /5

studysmarter schule studium
d

4.8 /5

Lerne jetzt mit Karteikarten und Zusammenfassungen für den Kurs Introduction to Deep Learning an der TU München.

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

What happens if we initialize the weights if big random numbers

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

A demographic dataset with statistics on different cities' population, GPD per capita, economic growth is an example of "unstructured" data because it contains data coming from different sources.

Wählen Sie die richtigen Antworten aus:

  1. False

  2. True

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

What optimization methods exist? Which one is the choice for neural networks?

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

What happens if we initialize all weights with small random numbers?

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

Which statement is true?

Wählen Sie die richtigen Antworten aus:

  1. The deeper layers of neural network are typically computing more complex features of the input than the earlier layers.

  2. The earlier layers of a neural network are typically computing more complex features of the input than the deeper layers.

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

Which ones are "hyperparameters"?

Wählen Sie die richtigen Antworten aus:

  1. bias vectors b[ˡ]

  2. number of layers L in the neural network

  3. learning rate α

  4. number of iterations

  5. weight matrices W[ˡ]

  6. size of the hidden layers n[ˡ]

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

During forward propagation, in the forward function for a layer l you need to know what is the activation function in a layer (Sigmoid, tanh, ReLU, etc.). During backpropagation, the corresponding backward function also needs to know what is the activation function for layer l, since the gradient depends on it.

Wählen Sie die richtigen Antworten aus:

  1. True

  2. False

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

You have built a network using the tanh activation for all the hidden units. You initialize the weights to relative large values, using np.random.randn(..,..)*1000. What wilil happen?

Wählen Sie die richtigen Antworten aus:

  1. This will cause the inputs of the tanh to also be very large, causing the units to be „highl activated“ and thus speed up learning compared to if the weights had to start from small values.

  2. This will cause the inputs of the tanh to also be very large, thus causing gradients to also become larg. You therefore have to set α to be very small to prevent divergence; this will slow down learning.

  3. IIt doesn’t matter. So long as you initialize the weights randomly gradient descent is not affected by whether the weights are large or small.

  4. This will cause the inputs of the tanh to also be very large, thus causing gradients to be close to zero. Thze optimization algorithm will thus become slow.

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

Logistic regression's weights w should be initialized randomly rather than to all zeros, because if you initialize to all zeros, then logistic regression will fail to learn a useful decision boundary because it will fail to "break symmetry".

Wählen Sie die richtigen Antworten aus:

  1. True

  2. False

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

Suppose you have built a neural network. You decide to initialize the weights and biases to be zero. Which of the following statements is true?

Wählen Sie die richtigen Antworten aus:

  1. Each neuron in the first hidden layer will perform the samme computation. So even after multiple iterations of gradient descent each neuron in the layedr will be computing the same thing as other neurons.

  2. Each neuron in the first hiden layer will perform the same computation in the first iteration. But after one iteration of gradient descent they will learn to compute different things because we have „broken symmetry“.

  3. Each neuron in the first hidden layer will compute the same thing, but neurons in different layers will compute different things, thus we have accomplished „symmetry breaking“ as described in the lecture.

  4. The first hidden layer’s neurons will perform different computations from each other even in the first iteration; their parameters will thus keep evolving in their own way.

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

You are building a binary classifier for recognizing cucumbers (y=1) vs. watermelns (y=0). Which one of these activation functions would you recommend using for the output layer?

Wählen Sie die richtigen Antworten aus:

  1. ReLU

  2. Leaky ReLU

  3. sigmoid

  4. tanh

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

What is the "cache" used for in our implementation of forward propagation and backward propagation

Wählen Sie die richtigen Antworten aus:

  1. It is used to cache the intermediate values of the cost function training.

  2. We use it to pass variables computed during backward propagation to corresponding forward propagation step. It containts useful values for forward propagation to compute activations.

  3. We use it to pass variables computed during forward propagation to the corresponing backward propagation step. It containts useful values for backward propagation to computer derivatives.

  4. It is used to keep track of the hyperparameters that we are searching over to speed up computation.

Kommilitonen im Kurs Introduction to Deep Learning an der TU München. erstellen und teilen Zusammenfassungen, Karteikarten, Lernpläne und andere Lernmaterialien mit der intelligenten StudySmarter Lernapp. Jetzt mitmachen!

Jetzt mitmachen!

Flashcard Flashcard

Beispielhafte Karteikarten für Introduction to Deep Learning an der TU München auf StudySmarter:

Introduction to Deep Learning

What happens if we initialize the weights if big random numbers

Everything is saturated

Introduction to Deep Learning

A demographic dataset with statistics on different cities' population, GPD per capita, economic growth is an example of "unstructured" data because it contains data coming from different sources.

  1. False

  2. True

Introduction to Deep Learning

What optimization methods exist? Which one is the choice for neural networks?

  • ‚Vanilla‘ SGD
  • Momentum
  • RMSProp
  • Adagrad
  • Adadelta
  • AdaMax
  • Nada
  • AMSGrad
  • Adam -> Method of choice for neural networks

Introduction to Deep Learning

What happens if we initialize all weights with small random numbers?

Gradients vanish

Introduction to Deep Learning

Which statement is true?

  1. The deeper layers of neural network are typically computing more complex features of the input than the earlier layers.

  2. The earlier layers of a neural network are typically computing more complex features of the input than the deeper layers.

Introduction to Deep Learning

Which ones are "hyperparameters"?

  1. bias vectors b[ˡ]

  2. number of layers L in the neural network

  3. learning rate α

  4. number of iterations

  5. weight matrices W[ˡ]

  6. size of the hidden layers n[ˡ]

Introduction to Deep Learning

During forward propagation, in the forward function for a layer l you need to know what is the activation function in a layer (Sigmoid, tanh, ReLU, etc.). During backpropagation, the corresponding backward function also needs to know what is the activation function for layer l, since the gradient depends on it.

  1. True

  2. False

Introduction to Deep Learning

You have built a network using the tanh activation for all the hidden units. You initialize the weights to relative large values, using np.random.randn(..,..)*1000. What wilil happen?

  1. This will cause the inputs of the tanh to also be very large, causing the units to be „highl activated“ and thus speed up learning compared to if the weights had to start from small values.

  2. This will cause the inputs of the tanh to also be very large, thus causing gradients to also become larg. You therefore have to set α to be very small to prevent divergence; this will slow down learning.

  3. IIt doesn’t matter. So long as you initialize the weights randomly gradient descent is not affected by whether the weights are large or small.

  4. This will cause the inputs of the tanh to also be very large, thus causing gradients to be close to zero. Thze optimization algorithm will thus become slow.

Introduction to Deep Learning

Logistic regression's weights w should be initialized randomly rather than to all zeros, because if you initialize to all zeros, then logistic regression will fail to learn a useful decision boundary because it will fail to "break symmetry".

  1. True

  2. False

Introduction to Deep Learning

Suppose you have built a neural network. You decide to initialize the weights and biases to be zero. Which of the following statements is true?

  1. Each neuron in the first hidden layer will perform the samme computation. So even after multiple iterations of gradient descent each neuron in the layedr will be computing the same thing as other neurons.

  2. Each neuron in the first hiden layer will perform the same computation in the first iteration. But after one iteration of gradient descent they will learn to compute different things because we have „broken symmetry“.

  3. Each neuron in the first hidden layer will compute the same thing, but neurons in different layers will compute different things, thus we have accomplished „symmetry breaking“ as described in the lecture.

  4. The first hidden layer’s neurons will perform different computations from each other even in the first iteration; their parameters will thus keep evolving in their own way.

Introduction to Deep Learning

You are building a binary classifier for recognizing cucumbers (y=1) vs. watermelns (y=0). Which one of these activation functions would you recommend using for the output layer?

  1. ReLU

  2. Leaky ReLU

  3. sigmoid

  4. tanh

Introduction to Deep Learning

What is the "cache" used for in our implementation of forward propagation and backward propagation

  1. It is used to cache the intermediate values of the cost function training.

  2. We use it to pass variables computed during backward propagation to corresponding forward propagation step. It containts useful values for forward propagation to compute activations.

  3. We use it to pass variables computed during forward propagation to the corresponing backward propagation step. It containts useful values for backward propagation to computer derivatives.

  4. It is used to keep track of the hyperparameters that we are searching over to speed up computation.

Melde dich jetzt kostenfrei an um alle Karteikarten und Zusammenfassungen für Introduction to Deep Learning an der TU München zu sehen

Singup Image Singup Image
Wave

Andere Kurse aus deinem Studiengang

Für deinen Studiengang Informatik an der TU München gibt es bereits viele Kurse auf StudySmarter, denen du beitreten kannst. Karteikarten, Zusammenfassungen und vieles mehr warten auf dich.

Zurück zur TU München Übersichtsseite

Blockchain

Cognitive System

Databases for modern CPU

Protein Prediction I

Data Analysis in R

18WS_Strategisches_IT_Management

Patterns

Softwaretechnik

Autonomous Driving

Requirements Engineering

Web Application Engineering

Business Analytics

Echtzeitsysteme

Patterns in Software Engineering

Principles of Economics

Data Mining and Knowledge Discovery

Data Mining and KD

Visual data analytics

Was ist StudySmarter?

Was ist StudySmarter?

StudySmarter ist eine intelligente Lernapp für Studenten. Mit StudySmarter kannst du dir effizient und spielerisch Karteikarten, Zusammenfassungen, Mind-Maps, Lernpläne und mehr erstellen. Erstelle deine eigenen Karteikarten z.B. für Introduction to Deep Learning an der TU München oder greife auf tausende Lernmaterialien deiner Kommilitonen zu. Egal, ob an deiner Uni oder an anderen Universitäten. Hunderttausende Studierende bereiten sich mit StudySmarter effizient auf ihre Klausuren vor. Erhältlich auf Web, Android & iOS. Komplett kostenfrei. Keine Haken.

Awards

Bestes EdTech Startup in Deutschland

Awards
Awards

European Youth Award in Smart Learning

Awards
Awards

Bestes EdTech Startup in Europa

Awards
Awards

Bestes EdTech Startup in Deutschland

Awards
Awards

European Youth Award in Smart Learning

Awards
Awards

Bestes EdTech Startup in Europa

Awards

So funktioniert's

Top-Image

Individueller Lernplan

StudySmarter erstellt dir einen individuellen Lernplan, abgestimmt auf deinen Lerntyp.

Top-Image

Erstelle Karteikarten

Erstelle dir Karteikarten mit Hilfe der Screenshot-, und Markierfunktion, direkt aus deinen Inhalten.

Top-Image

Erstelle Zusammenfassungen

Markiere die wichtigsten Passagen in deinen Dokumenten und bekomme deine Zusammenfassung.

Top-Image

Lerne alleine oder im Team

StudySmarter findet deine Lerngruppe automatisch. Teile deine Lerninhalte mit Freunden und erhalte Antworten auf deine Fragen.

Top-Image

Statistiken und Feedback

Behalte immer den Überblick über deinen Lernfortschritt. StudySmarter führt dich zur Traumnote.

1

Lernplan

2

Karteikarten

3

Zusammenfassungen

4

Teamwork

5

Feedback