Questions tagged [deep-learning]
Deep Learning is an area of machine learning whose goal is to learn complex functions using special neural network architectures that are "deep" (consist of many layers). This tag should be used for questions about implementation of deep learning architectures. General machine learning questions should be tagged "machine learning". Including a tag for the relevant software library (e.g., "keras", "tensorflow","pytorch","fast.ai" etc) is helpful.
deep-learning
27,896
questions
-1
votes
0
answers
22
views
How exactly would I write the validation loop for mobilenet object detection in PyTorch?
I'm training a MobileNet V3 Small model for object detection using PyTorch. I've successfully implemented the training loop, but I'm unsure how to write the validation loop correctly. I need help with ...
- 1
-1
votes
0
answers
10
views
About time series data prediction model dataset using lstm model
I have average traffic volume data by year for 1400 locations, and I have data from 2008 to 2022. At this time, in order to conduct time series analysis and create a prediction model, we are planning ...
- 1
-2
votes
0
answers
26
views
How to parallelize the convolutional layers of Darknet53 of yolov5 (on one PC)? [closed]
I want to split the backbone algorithm of YOLOv5 and deploy it on multiple FPGAs, but first I need to divide the Darknet53 convolution layer into 2 or more regions on a computer for parallel ...
-2
votes
0
answers
23
views
How to make adapter_conditioning_scale trainable in multiple T2I_Adapter? [closed]
Below is the code using Multi T2I_Adapter. As seen in the code below, adapter_conditioning_scale=[0.8, 0.8], is manually set.
adapters = MultiAdapter(
[
T2IAdapter.from_pretrained("...
- 49
-5
votes
0
answers
21
views
deep learning, library, coding [closed]
In the current deep learning support libraries such as tensorflow hub, pytorch torch, hugging face transformer, there are many built-in functions such as tokenizer etc.,
how can I know what these ...
-2
votes
0
answers
14
views
MLP Regressor Engineering Data SKLearn
I have 10 accelerometers distributed on an aircraft analytical model. From my analytical model, I have a set of sensor acceleration of 10 Accelerometers X 6 Degrees of Freedom X 6000 (60 seconds) data ...
- 1
-3
votes
0
answers
18
views
Forecasting values with a Neural Network [closed]
I am working with a neural network for oxygen time series regression (TensorFlow and Keras), I have oxygen, saturation, temperature and salinity data every 5 minutes from 2023-08-25 00:02:12 to 2024-...
0
votes
0
answers
29
views
The Impact of Pretraining on Fine-tuning and Inference
I am working on a binary prediction classification task, primarily focusing on fine-tuning a BERT model to learn the association between CVEs and CWEs. I've structured my task into three phases: first,...
- 19
-1
votes
0
answers
5
views
Keras tuner raising RuntimeError
Whenever I want to tune hyperparameter with keras tuner it raises exception saying RuntimeError: Number of consecutive failures exceeded the limit of 3. Can anyone help me how to solve this problem?
I ...
-2
votes
0
answers
10
views
LSTM could not broadcast input array from shape (2302,) into shape (1440,) [closed]
ValueError Traceback (most recent call last)
Cell In[59], line 114
111 print(len(test_predict))
113 # Assign predictions to the respective plot segments
114 ...
0
votes
0
answers
24
views
Unidentified TensorFlow retracing leading to ResourceExhaustedError
I'm using TF Keras to build and train two ML models together and I'm new to it. My Encoder below is a U-Net CNN that aims to invisibly embed a static watermark onto a cover image.
class Encoder(Layer):...
- 1
0
votes
0
answers
15
views
Segment Anything Model (SAM) How do i predict_torch with multiple boxes with their corresponding points?
I am trying out Segment Anything Model (SAM) and my problem needs a multiple boxes with their corresponding points for specificity inside the box. for example box1 = [#, #, # ,#] with its points [x,y] ...
- 15
0
votes
0
answers
17
views
There seems to be something wrong with my code implementation of the evaluation metric ndcg
def ndcg(preds: dict,test_gd: dict, topN=50):
"""NDCG@topN
Args:
preds (dict): preds[user] = [item1,item2,...]
test_gd (dict): test_gd[user] = [item1,item2,...]
...
- 9
-1
votes
0
answers
32
views
How to use Intel Xe Graphics integrated GPU to train deep learning pytorch models? [closed]
I have an integrated GPU "Intel(R) Iris(R) Xe Graphics" in my device and i want to use it to train my deep learning model build in pytorch.
I tried to install Intel oneAPI Base Toolkit and ...
- 1
0
votes
1
answer
19
views
Depreciation of some function in python: Ambiguous truth value of dataframe
I was training various ANN when I encountered this error in this part. I don't know if any function has been depreciated or not.
def analyze_results(rainfall_data, test_rainfall_data, name, flag=False)...
-3
votes
0
answers
24
views
How can i resolve the preblem? [closed]
I defined a model and then initialized another model within it. However, I did not use this initialized model. It still affects my results. Why is that?
this is the model i do not use but initialized ...
- 5
1
vote
1
answer
30
views
Broadcasting multiple version of X_data that pairing with same y_data
My deep learning architecture accept input vector with size 512 and output vector with size 512 too.
The problem is I have X_data version that pairing with same y_data.
I have this tensors:
(4, 8, 512)...
-4
votes
0
answers
26
views
Neural Networks take too long to train - I'm exploring renting GPUs that can speed up time taken to run [closed]
I've running various machine learning problems to predict prices (in financial markets). I've had some success using tree based algorithms. However I want to also test deep learning neural nets - but ...
-2
votes
0
answers
18
views
unable to deploy my deep learning app based on flask [closed]
my ml app based out of flask,it takes pdf as input and lets users ask questions and utilizes BERT model to answer it,
so far it is running locally good but no luck in deployment i tried vercel, ...
- 1
-2
votes
0
answers
15
views
Time-serie outlier detection using LLM
I want to preform outliers detection on time-serie data using an LLM (the use of LLM is critical). so basically i have a simple dataframe containing the time serie data (timestamp column, and value ...
-1
votes
0
answers
20
views
Train a PINN to invert for unknown parameters
I'm using a PINN to solve the damped oscillator differential equation and simultaneously find the friction parameter of the latter, given noisy observations of a damped oscillator as an input. I wrote ...
- 1
-1
votes
0
answers
13
views
How to use the Vgg16 optimally
I read some code on the Vgg16 and I have some questions please. In some code I notice that they use the outputs like this :
vgg = VGG19(weights="imagenet",include_top=False, input_shape=...
- 39
-1
votes
1
answer
34
views
Vision Transformers (ViT) - Clarifying question
I found below code snippet online and I got basic image classification training working on google colab.
Can someone explain in simple terms what exactly is happening under the hood when we do ...
- 49
-1
votes
0
answers
29
views
How can I solve the UnicodeDecodeError? [closed]
I tried to train faster rcnn with my dataset but i always get this error. "UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfd in position 118: invalid start byte".
This is the ...
0
votes
0
answers
28
views
The "accelerate" package version not compatible with transformers library
I'm going through the Huggingface tutorial. In this section, I met a problem with this code:
from transformers import TrainingArguments
training_args = TrainingArguments("test-trainer")
...
0
votes
0
answers
9
views
"Can't load feature extractor for 'facebook/wav2vec2-large-960h-lv60-self" Error when I try to extract Audio Features
The Code
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")
model = Wav2Vec2Model.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")
...
0
votes
0
answers
18
views
CRNN OCR, issue with dynamic width of images. How to tackle and resolve this? Need to train on images with different width
I am creating a ocr for urdu. I bucket my data into n number of buckets based on their widths. means all images whose widths are less than 400px are all reshaped to have width of 400px and are stored ...
- 19
0
votes
0
answers
31
views
Serious delays in Keras before training the neural network
I'm creating a NN that has 2 levels.
At the first level, there are several small neural networks that should have a good generalizing ability for the data being fed to them. At the second level there ...
-2
votes
0
answers
9
views
CNN Predicting Random Images
I have Trained the model with CNN. The MODEl i have Trained for Binary classification with chest - X - Ray images. Even after training the model. Model is predicting Random images like cars, animals ...
1
vote
0
answers
29
views
problem on state representaion Finrl StockTradingEnv
i'm trying to adjust and apply my own trading strategies on Finrl meta environment and ensmeble agent method. in this process i decide to add some new features which are produced by a neuran network ...
- 11
-1
votes
0
answers
50
views
Image stitching using Poisson Blending
Background
I'm working on an image stitching project using the Poisson blending method. I found a repository on GitHub that deals with the same problem (https://github.com/erman18/poison-blending-...
- 7
-2
votes
0
answers
22
views
Issues about CIFAR-10 and transfer learning [closed]
I've tried many and many times, and it doesn't work at that line:
history = model.fit(train_generator, epochs=200, steps_per_epoch=167,
callbacks=[modelcheckpoint],
...
1
vote
1
answer
46
views
How do I define/change the accuracy for a non-classification convolutional neural network?
I'm using Keras to make a prediction model. It takes in two time series and outputs a number between 0 and 1. Currently, I am getting very low accuracy as the model is only considered "correct&...
- 21
0
votes
0
answers
35
views
My tensor values in the prediction are too low
I'm training a semantic segmentation model, using PyTorch and the U-net architecture. The dataset consists of retinography images, with masks of exudate diseases. And I have a problem with ...
0
votes
0
answers
31
views
Fine-tuned Mask2Former model only returning null-tensors after training
I have written code to fine-tune a mask2former model. However when running inference on it, it always just returns a null-tensor.
Now the model gets trained with a (384, 384) RandomCrop of a (1080, ...
- 11
-1
votes
0
answers
27
views
How to Improve Tree Detection and Counting in Forest Satellite Imagery Using Pretrained Models? [closed]
I am developing an AI-powered forestry management system to monitor and analyze tree populations using satellite imagery from Google Earth. The primary goal is to count trees accurately and identify ...
-2
votes
0
answers
25
views
I have a python problem in case of data model predictions [closed]
I have a block of codes that isnt working as it should, after traning the datasets and doing all that is required my face mask detection system isnt detecting when user wear mask
I tried using ...
0
votes
0
answers
9
views
how is the derivative of the loss function wrt to the inputs in final layer equals to y_true/dvalues where dvalues is the dervative wrt to output?
while reading about neural network i came across
`class Loss_CategoricalCrossentropy(Loss):
# Backward pass
def backward(self, dvalues, y_true):
# Number of samples
samples = len(dvalues)
# Number of ...
- 1
0
votes
0
answers
14
views
Why reshape is needed for Sequential models after minmaxscaler class? and how to use?
My recent project is based on 'Stock Prediciton" I'm working on df["Open"] as nd.array.
Trying to understand intuitively how this works and whether my predicted data will be correct or ...
0
votes
0
answers
14
views
What's the problem with my tf.keras.input() function
when trying to train my model i am getting error - "ValueError: Cannot convert '8' to a shape", when the shape of my X_train is (5347,28).
import tensorflow as tf
from tensorflow.keras ...
-1
votes
0
answers
19
views
My ConvLSTM model for predicting sign language gestures trained on WLASL video dataset is not improving accuracy during training phase for 10 epochs
Link for the dataset - https://www.kaggle.com/datasets/risangbaskoro/wlasl-processed
This is preprocess dataset code.
# IMPORTS
import os
import sys
import pandas as pd
import torch
from torch.utils....
-1
votes
0
answers
9
views
the loss in GNN doesn't decrease
When I'm studying the GNN,my training code can't train the module and the parameter keep still
def create_node_emb(num_node=34, embedding_dim=16):
# TODO: Implement this function that will create ...
-2
votes
0
answers
24
views
Should I train an OCR model for handwritten text recognition for a mobile app or just avail API services?
Is it worth building a handwritten text recognition model for a mobile app (need to finish this project by next month)? Must be at least 75% accurate.
I'm building a simple Android mobile app (react ...
- 1
0
votes
0
answers
47
views
Parsing Error with YOLOv3 and OpenCV 4.10.0 when reading yolov3.cfg
I'm encountering a parsing error when attempting to load a YOLOv3 model using OpenCV 4.10.0 in Python. Specifically, the error occurs while trying to read the configuration file yolov3.cfg. Despite ...
0
votes
0
answers
40
views
How to access or use PaddleOCR API for handwritten text purposes (this is for my mobile application)?
I'm a beginner building an Android mobile app which has a feature of detecting handwritten texts. I noticed that one of the good models available and is for free is the PaddleOCR. However, the Github ...
- 1
0
votes
0
answers
35
views
Rewards not increasing in Actor-Critic model in a custom environment
I am trying to implement an Actor-Critic Deep Reinforcement Learning model to predict training intensities for clients with different processing powers in a distributed machine learning setup (...
-1
votes
0
answers
29
views
How do the top and bottom layers of AlexNet communicate? [closed]
I am trying to re-implement Krizhevsky et al. (2012) using PyTorch, and I am confused how precisely the second and the third convolutional layers of the AlexNet model communicate (and same for inputs ...
-1
votes
0
answers
24
views
RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x1 and 35x256)
I keep getting this error and I'm not sure how to fix it. I'm working on a PPO code and this is my first time. The error is in the mu = self.actor(state) and choose_action function.
I asked ChatGPT ...
- 1
-1
votes
0
answers
28
views
Word Prediction to complete sentence [closed]
The visuals in this movie are _________, but the story is paper-thin.
It's all flash and no substance.
How can I better recommend word to complete the sentence
Tried a code and got output as The ...
- 7
0
votes
1
answer
33
views
Loss function not decreasing on a CNN model?
I am a creating a CNN model in order to detect emotions (I am very new to creating neural networks). The dataset I used had this structure:
DatasetName ->
train ->
0
1
2
3
4
5
6
7
where each ...
- 1