Unanswered Questions
3,731 questions with no upvoted or accepted answers
13
votes
1answer
195 views
Will parameter sweeping on one split of data followed by cross validation discover the right hyperparameters?
Let's call our dataset splits train/test/evaluate. We're in a situation where we require months of data. So we prefer to use the evaluation dataset as infrequently as possible to avoid polluting our ...
12
votes
3answers
439 views
Can some one help me understand this paragraph from Nvidia's progressive GAN paper?
Furthermore, we observe that mode collapses traditionally
plaguing GANs tend to happen very quickly, over the course of a dozen mini-batches. Commonly
they start when the discriminator overshoots, ...
11
votes
2answers
240 views
Why does Batch Normalization work?
Adding BatchNorm layers improves training time and makes the whole deep
model more stable. That's an experimental fact that is widely used in machine
learning practice.
My question is - why does it ...
10
votes
0answers
166 views
What is the number of neurons required to approximate a polynomial of degree n?
I learned about the universal approximation theorem from this guide. It states that a network even with a single hidden layer can approximate any function within some bound, given a sufficient number ...
8
votes
0answers
92 views
Why is my GAN more unstable with bigger networks?
I am working with generative adversarial networks (GANs) and one of my aims at the moment is to reproduce samples in two dimensions that are distributed according to a circle (see animation). When ...
8
votes
1answer
230 views
Extending FaceNet’s triplet loss to object recognition
FaceNet uses a novel loss metric (triplet loss) to train a model to output embeddings (128-D from the paper), such that any two faces of the same identity will have a small Euclidean distance, and ...
8
votes
2answers
328 views
Back-of-the-envelope machine learning (specifically neural networks) calculations
There is a popular story regarding the back-of-the-envelope calculation performed by a British physicist named G. I. Taylor. He used dimensional analysis to estimate the power released by the ...
7
votes
1answer
159 views
How to deal with a small amount of labeled samples?
I'm trying to develop skills to deal with very small amounts of labeled samples (250 labeled/20000 total, 200 features) by practicing on Kaggle "Don't Overfit" dataset (Traget_Practice have ...
7
votes
0answers
896 views
Is there a difference in the architecture of deep reinforcement learning when multiple actions are performed instead of a single action?
I've built a deep deterministic policy gradient reinforcement learning agent to be able to handle any games/tasks that have only one action. However, the agent seems to fail horribly when there are ...
7
votes
2answers
156 views
Why are documents kept separated when training a text classifier?
Most of the literature considers text classification as the classification of documents. When using the bag-of-words and Bayesian classification, they usually use the statistic TF-IDF, where TF ...
7
votes
1answer
126 views
Using ConceptNet5 to find similar systems to solve specific problems?
I installed a local running instance of the ConceptNet5 knowledgebase in an elasticsearch server. I used this data to implement the so-called "Analogietechnik" (a creativity technique to solve a ...
6
votes
0answers
170 views
Is the Bellman equation that uses sampling weighted by the Q values (instead of max) a contraction?
It is proved that the Bellman update is a contraction (1).
Here is the Bellman update that is used for Q-Learning:
$$Q_{t+1}(s, a) = Q_{t}(s, a) + \alpha*(r(s, a, s') + \gamma \max_{a^*} (Q_{t}(s',
...
6
votes
2answers
188 views
How can I solve the zero subset sum problem with hill climbing?
I want to solve the zero subset sum problem with the hill-climbing algorithm, but I am not sure I found a good state space for this.
Here is the problem: consider we have a set of numbers and we want ...
6
votes
0answers
60 views
Normalizing Normal Distributions in Thompson Sampling for online Reinforcement Learning
In my implementation of Thompson sampling (TS) for online Reinforcement Learning, my distribution for selecting $a$ is $\mathcal{N}(Q(s, a), \frac{1}{C(s,a)+1})$ where $C(s,a)$ is the number of times $...
6
votes
1answer
56 views
What is the impact of using multiple BMUs for self-organizing maps?
Here's a sort of a conceptual question. I was implementing a SOM algorithm to better understand its variations and parameters. I got curious about one bit: the BMU (best matching unit == the neuron ...