Questions tagged [deep-learning]

For questions related to deep learning, which refers to a subset of machine learning methods based on artificial neural networks (ANNs) with multiple hidden layers. The adjective deep thus refers to the number of layers of the ANNs. The expression deep learning was apparently introduced (although not in the context of machine learning or ANNs) in 1986 by Rina Dechter in the paper "Learning while searching in constraint-satisfaction-problems".

Filter by
Sorted by
Tagged with
0 votes
0 answers
20 views

Exploring the Similarity of Sibling’s Voices Using Automatic Speaker Recognition

I want to start project on Exploring the Similarity of Sibling’s Voices Using Automatic Speaker Recognition Everyone has a unique voice, because of the different structure of their articulatory ...
Alan Turing's user avatar
0 votes
0 answers
12 views

Possible Reasons for the Discrepancy in Trainable Parameters of the Extended DeepConvLSTM Model

I have implemented DeepConvLSTM baseline Model input are 60×d frames each representing 60 samples with d features. Frames are fed into four consecutive convolution layers with standard rectified ...
Nafees Ahmed's user avatar
1 vote
1 answer
130 views

Reinforcement Learning vs Supervised Learning [duplicate]

I have never tried reinforcement learning in my life. I'm planning to apply it in robotics. I have some experiences using supervised learning mainly deep learning. So, that's mean I will use neural ...
Muhammad Ikhwan Perwira's user avatar
1 vote
0 answers
25 views

Choosing a Deep Learning model to analyse microscope images

I have lots of simulated training data of microscope images and I want to train a network to count the number of points in the image. These points are distributed in concentric oblate shells. The ...
James's user avatar
  • 111
1 vote
0 answers
27 views

Is it possible to have a good validation accuracy with random labels?

I'm currently trying to train a siamese network to determine if two inputs are similar or not. The inputs are power consumption traces and I'm basically using the siamese network as some sort of ...
WINTERSDORFF Raphael's user avatar
1 vote
0 answers
36 views

Siamese network, cosine similarity unexpected result?

I was reading more about siamese network and it's use for similarity problems and I've stumbled upon this https://keras.io/examples/vision/siamese_network/ I was surprised to see both similarities in ...
recimo's user avatar
  • 11
0 votes
1 answer
175 views

How to force Transformer to give more weight to certain tokens

I'm developing an encoder-decoder based transformer model and I would like to ask if there are ways to incentivize or penalize certain tokens during training. I'm working on a translation task where ...
jasperagrante's user avatar
0 votes
0 answers
47 views

How does a zero-order hold kernel in a Convolutional Neural Network look like?

Several papers co-authored by Hitoshi Kiya propose to use a fixed convolutional layer with a zero-order hold kernel to avoid checkerboard artifacts in CNNs. [1, 2, 3] While there is plenty of ...
Domderon's user avatar
  • 101
1 vote
1 answer
31 views

LSTM with multiple data streams

I am working on the following problem: I have ~10 weather stations in somewhat approximate areas, at some points during the day (different for each station), I get readings of various data points (...
PenguinHook's user avatar
4 votes
5 answers
962 views

Why does Stable Diffusion use VAE instead of AE?

I am currently studying the Latent Diffusion Models (LDMs) and am interested in training my own model using a unique dataset. In my research, I came across Stable Diffusion (SD). Some sources suggest ...
P0TAT0's user avatar
  • 41
1 vote
1 answer
63 views

Which process is better to understand images?

What is the difference between this process of recognizing objects in a image: (The correlation function calculate the correlation coefficient between the input and a image containing the object we ...
Cerise's user avatar
  • 121
0 votes
0 answers
59 views

RAM usage increases linearly while training. What could be causing the issue?

I'm training a siamese neural network with semi-hard batching, RAM usage increases linearly over epochs, peaking at 60GB, then switches from GPU to CPU, slowing down training. I'm using Windows with ...
Ahmed Altunkaya's user avatar
0 votes
1 answer
47 views

Gradient: any resource on how to understand everything about it?

I have read some resources about AI, and they all speak about the gradient. Is there any book focused on this? maybe with tons of images / diagrams? Cheers
zerunio's user avatar
0 votes
0 answers
50 views

How to use deep learning to train an AI to play a two-player tactic game like rock-papers-scissors?

I want to try to train an AI model that can play a simple two-player tactic game. The game is not rock-papers-scissors, but have similar properties. Firstly, two players present their moves ...
Jason Jia's user avatar
0 votes
0 answers
7 views

autoencoders for anomaly detection, training individual models for different users or roles, how?

Do I first train a generic model for all of my users on a network, say for a network anomaly detection example, then fine tune for each user on their own subset of the training data? But I'd be using ...
mLstudent33's user avatar
0 votes
0 answers
20 views

Convert specific domain knowledge text to a knowledge graph

As part of this semester assignment , I'm working on a project that aims to to represent the knowledge in "PMBOK 6th edition, section 11: Project Risk Management (page 395 -> 458)" and the knowledge ...
Wissem Boujlida's user avatar
0 votes
0 answers
14 views

Build a global model based on thousand local models

I have 1000 meters for electricity usage. I want to train one global model for all of them to predict their consumption for the next few days. So, when I train the model with all of the meter data, ...
Sadcow's user avatar
  • 99
0 votes
0 answers
28 views

What causes train loss to suddenly rise, and never fall back, starting at some epoch?

Please see the following train/val losses of some basic vision binary classifier training I am performing. I thought train loss should always go down. When dropping the learning rate, this still ...
Gulzar's user avatar
  • 759
1 vote
0 answers
58 views

How does diffusion model (DDPM) ensures novel generated samples?

I am trying to understand the theoretical aspect of the denoising diffusion model. There we try to destroy the initial image x_0 through a chain of forward process and then learn a backward diffusion ...
Formal_this's user avatar
4 votes
0 answers
87 views

Why policy gradient theorem has two different forms?

I have been studying policy gradients recently but found different expositions from different sources, which greatly confused me. From the book "Reinforcement Learning: an Introduction (Sutton &...
Yuxiang Wei's user avatar
0 votes
0 answers
32 views

Approaches for multi-label classification with over 1,000,000 labels

I have billions of rows in some dataset and each row can be in any subset of about 1 million binary labels. So the number of overall classes would be $\sim 2^{1,000,000}$, if I were to think about it ...
economicagent's user avatar
0 votes
0 answers
8 views

Algorithm Suggestion for Diverging Data (Severity/Intensity Analysis)

I have four datasets for four different accidents; each dataset has the same parameters. Some of the key parameters are changing their values from a "standard value". The more they change ...
Rubayet Alam's user avatar
0 votes
0 answers
21 views

ROC curve for multiclassification - results sound not correct

I'm working on a multiclassification task using LSTM algorithm, i generated my roc curve plots but they give scores like 1 , 0.99, 0.97 however i have an accuracy of 0.97, Precision 0.65, Sensitivity/...
biihu's user avatar
  • 1
0 votes
0 answers
32 views

Pixel-wise regression only focus on edge

I am trying to use unet to learn pixel-wise regression from one image to one groundtruth with the same image size. The network seems to focus too much on the edge of the image, and it does not learn ...
K.Nguyen's user avatar
0 votes
1 answer
57 views

Why Does the Model not Improve in PyTorch?

I have a simple curve fitting problem in hand. I wrote some code in PyTorch as follows: ...
Burak Karaosmanoğlu's user avatar
0 votes
0 answers
35 views

MultiLayer Perceptron not working for regression problem, what could I try?

I am trying to learn the inverse kinematics of a robotic manipulator. To do that I have a simulator with which I acquired data. My dataset is composed of positions in X, Y and Z and actuator variables ...
Guillaume's user avatar
1 vote
0 answers
24 views

Toy dataset: Radial VAE

I'm evaluating disentanglement in toy datasets seeing as we have such little understanding of the phenomena. I'm using various tools from differential geometry. Now I want to train a VAE on the ...
John Miller's user avatar
0 votes
0 answers
25 views

Does splitting a model across multiple GPUs (using model parallelism) reduce model accuracy?

When training a model in PyTorch on multiple GPUs, it should ideally give the same results as using just one GPU. However, are there any specific things to be careful about during implementation?
willtryagain's user avatar
0 votes
0 answers
24 views

Yolo object detection metrics

I have made some predictions and saved the results to YOLO format. Then I made a program to calculate metrics, every metric looks fine except Precision/Confidence curve. I guess the flatline at the ...
Andrés Rodríguez Lorenzo's user avatar
0 votes
0 answers
153 views

How to implement Image Fusion Equation in Python

I am new to image processing. I am trying to implement this paper. (official implementation is available on github the link for the same is given below) I have applied one fusion strategy. I am not ...
programmer_04_03's user avatar
7 votes
4 answers
4k views

Do neural network weights need to add up to one?

The idea that weights determine how much influence each input value from the current layer will have when calculating the input to the following layer reminds me of when my professors would say that ...
Garrett's user avatar
  • 181
2 votes
2 answers
73 views

How translation invariance is achieved in CNNs?

I am trying to understand how translation invariance is achieved in CNNs. For example, consider the following simple binary classification problem: predicting whether the letter that appears on an ...
ado sar's user avatar
  • 150
1 vote
1 answer
91 views

How to transform a loss function into a score function?

Loss_Function/Maximize_Function/Score_Function, CustomLoss, pytorch. Using Custom Loss for Maximizing Score in PyTorch I'm using a PyTorch model with an LSTM input layer, a linear hidden layer, and 3 ...
IAQuestions's user avatar
1 vote
2 answers
146 views

How to generate a 3D model from only 1 image?

I'm posting this on the AI stack exchange because even though this can be solved with a "regular" complex and sophisticated algorithm, it seems that trying to generate something for which ...
OGOG's user avatar
  • 11
0 votes
0 answers
20 views

Can we use bounding box cropping to avoid shortcut learning (achieve explainable AI)?

Deep neural networks sometime use shortcut features (pseudo correlation) to predict. For example, in cat-dog classification, the network may use the background information (e.g. floor, grass) as a ...
Mingzhou Liu's user avatar
0 votes
0 answers
7 views

Mechanism of Prediction Readjustment in Supervised Learning and Role of Self-Attention in Sequence Data Relationships

In supervised learning, when the prediction deviates significantly from the expectation, how does it "readjust"? And... LLMs are a subset of deep learning, just as generative AIs are. Is the ...
Assandra Lakal's user avatar
0 votes
0 answers
32 views

Which meta-learning algorithms are well-suited for "many-shot learning" scenarios, where the target training set is large?

Much of the meta-learning literature deals with the few-shot learning problem of using data from a diverse set of "source" tasks (the meta-dataset) in order to train a model that can quickly ...
Ori's user avatar
  • 101
1 vote
1 answer
295 views

What causes my loss curve to consistently oscillate when training an LLM?

Why is my loss curve consistently oscillating? Every 50 steps it jumps back up. I'm assumming there's a bug in my data, since I'm using this colab notebook that shows a proper train/loss at the bottom....
bmatzelle's user avatar
0 votes
0 answers
5 views

Multi Output Regression but num of objects to predict vary per sample

so recently I came across a problem of predicting the positions of objects from a pulse wave. My biggest concern here is that for each data sample, the number of objects varies. I know that this ...
mim96's user avatar
  • 1
0 votes
0 answers
18 views

Multi-instance learning for time-spatio-dependent data

I am trying to use MIL approach from the paper Attention-based Deep Multiple Instance Learning on the data that is a frames of human pose images acquired on different angle on each timepoint (temporal ...
PatrickHellman's user avatar
2 votes
1 answer
98 views

Why can Variational Autoencoders (VAEs) approximate arbitrary distributions?

I am trying to reason to myself why is it that VAEs can approximate arbitrary probability distributions even though 𝑞𝜙(𝑧|𝑥) and 𝑝𝜃(𝑥|𝑧) are Gaussian. I understand that the parameters are ...
Joel's user avatar
  • 33
1 vote
2 answers
96 views

Maximize a scoring function within the latent space of a generative model

Given a generative model, G, trained on a dataset D. This generative model can be either GAN or Diffusion based. Supposed each sample, x_i, generated by G, can be evaluated by a readily available ...
terenceflow's user avatar
1 vote
1 answer
42 views

How to handle large dimensionality differences between state and action inputs in a reinforcement learning predictor?

I'm currently writing code for a reward predictor function r=f(s,a) in reinforcement learning, where 's' is the state with 256 dimensions (the embedding dimension after visual input is processed by an ...
XiaoBanni's user avatar
4 votes
2 answers
184 views

Why are some Neural Networks more forgiving on Quantization?

I know this might be a bit general question and concerning a rather active research field, much beyond my expertise, but I do believe there're some answers. The use of NN parameters quantization can ...
edmz's user avatar
  • 91
1 vote
0 answers
39 views

Can back-bone of text-to-image GEN AI models utilised for classification?

With the advent of GEN AI (Stable Diffusion), we are able to create images with text. For eg. If i need to create a dog on beach during sunset; now in background this model needs to first get images ...
prat__'s user avatar
  • 33
0 votes
0 answers
19 views

How to Train Multiple Deep Learning Models on Multiple GPUs?

I have access to a GPU server with four gpus. Now I would like to train multiple models or folds of one model, one on each gpu. How can I schedule multiple trainings, so that a new training instance ...
e.Fro's user avatar
  • 186
0 votes
0 answers
66 views

How to tell an agent that some actions in the action space are currently not available in gym and the design of action space

I want to make a task allocation decision by reinforcement learning. Suppose there are N tasks to be allocated and M severs to serve these task. However, there is a constraint that one task should be ...
Reese's user avatar
  • 1
0 votes
0 answers
24 views

Binary classification using softmax and categorical crossentropy: monitoring validation

I run a binary classification using different CNN versions in Tensorflow. When I label samples from each class using 0 and 1, I select a sigmoid output in the last layer of the CNN, like ...
GKH's user avatar
  • 101
0 votes
0 answers
21 views

How to make better modelling with less data?

I was working on modelling with less data which is the one of the most problem on DL. So i learned somethings about to deal with it. The model was always overfitting. And i learned Data Augmentation. ...
Joseph's user avatar
  • 1
1 vote
1 answer
80 views

Why the gradients produced by the soft targets scale as 1/T^2 in knowledge distillation?

In the paper "Distilling the knowledge in a neural network", it mentioned "the magnitudes of the gradients produced by the soft targets scale as 1/(T^2) ", but it has no ...
Xiong's user avatar
  • 11

1 2
3
4 5
41