56,223 questions
-4
votes
0
answers
27
views
Why is my face recognition model detecting downloaded and live camera images but not detecting screenshots correctly, how to resolve this error? [closed]
Small Info About the Project:
This project is a Face Recognition System that can analyze images from both saved image files and webcam input. The model detects whether a given image is from a live ...
-1
votes
0
answers
9
views
Output layers in Mediapipe RTLite object detection model
I'm somewhat new to ML/AI and I trained my custom model exactly per the sample example given by mediapipe: https://colab.research.google.com/github/googlesamples/mediapipe/blob/main/examples/...
- 1
-1
votes
0
answers
21
views
I'm having an Error on hugging face and nothing is working
I'm trying to deploy a project on huggingface but there is an error that I can't seem to solve.
I'm getting this error
Cannot start Node server on any port in the range 9001-9001.
Please install Node ...
0
votes
1
answer
39
views
Inconsistent results when training models using different GPUs
I've been trying to train a language model (text classification), our lab has two GPUs, a 4090 and a 3090. However, I encountered a perplexing phenomenon during training: the model's performance ...
- 1
0
votes
1
answer
57
views
XGBoost does not predict properly on input that's equal to traning data
Why this quite simple example of XGBoost ML produces all-nulls even on input, that's equivalent to training data? This looks like a trivial case of input which should not require any fine tuning of ML,...
- 145
0
votes
0
answers
24
views
Failed to generate image embeddings: The size of tensor a (1246) must match the size of tensor b (77) at non-singleton dimension 1
I am using a sentence-transformers model to make embeddings of image files (PIL ImageFile). However, it gives the error in the title. I tried a bunch of things to try to solve it, but to no avail.
I ...
0
votes
0
answers
23
views
getting the text and tokens using layoutlmv3
I trained a layoutlmv3 model creating a label dataset using label studio. i was able to test the output of the model using the following code
encoding = processor(image, words, boxes=boxes, ...
- 145
-3
votes
0
answers
28
views
Fine-tuning LLAMA for small datasets [closed]
I'm looking to fine-tune LLAMA model to answer certain questions. Currently the knowledge base from which the questions must be answered is small, hence I can provide it as context and ask LLAMA to ...
0
votes
0
answers
17
views
exporting yolo segmentation model to coreml
I’m exporting the model like this:
model = YOLO('YOLO11m-seg.pt')
model.export(format="coreml")
And then loading into Xcode. Works great. Here's how I'm doing inference and inspecting the ...
-1
votes
0
answers
16
views
How to formulate semantic map as input? [closed]
I have a problem that requires semantic map of the scene to be passed through a Unet encoder. I wanted to know how to shape this map for input.
Right now I have a single channel map that has integer ...
0
votes
0
answers
22
views
Android mediapipe second inference instance can't be initialized
I do some experiments with google.mediapipe framework and observing next issue... if some instance/task was initialized, it's impossible to create another instance of inference task after.
For example,...
- 1,500
-1
votes
0
answers
15
views
How to Adapt Shapley Value Calculation for Sequential Attribution Models with Repeated Channels? [closed]
I'm developing an attribution model based on the Shapley Value to analyze marketing journeys. In my dataset, a single channel (e.g., "email", "phone call") can appear multiple ...
-4
votes
0
answers
46
views
Overfitting model: Difference between training accuracy and validation accuracy is large [closed]
I am building a geospatial model that can predict crime pattern for future years. I am training my model with kernel density raster data from 1999 - 2023. I created the raster from point feature ...
1
vote
0
answers
50
views
Warning when loading the model because of Adam optimizer
I am implementing the following multi-task model with the following layers:
@tf.keras.utils.register_keras_serializable()
class Bcb_block(Layer):
def __init__(self, mid_chan, out_chan, data_format ...
- 63
0
votes
0
answers
18
views
alternating epochs are resulting into all 0 results , loss and accuracy all zero results [closed]
while i am running my training loop i am getting a valid accuracy on one epoch but on every alternate epoch i am getting all 0 results my loss and accuracy all the results are 0
I was trying to train ...
- 29
-1
votes
0
answers
30
views
how tensorflow actually treat large/small logits [closed]
In the most binary classfication, we compute cross entropy as following
loss = y*log(p) + (1-y)log(1-p)
where p = 1/(1+exp(-logit))
And we all know when p->1 or p->0, loss actually explodes and ...
- 628
1
vote
1
answer
52
views
'numpy.ndarray' object has no attribute 'groupby'
I am trying to apply target encoding to categorical features using the category_encoders.TargetEncoder in Python. However, I keep getting the following error:
AttributeError: 'numpy.ndarray' object ...
- 11
-4
votes
0
answers
24
views
Any good datasets we can use to measure the performance of a personalized learning agent - trained on a Subject material? [closed]
I am planning to build a personalized learning agent, trained on an open source LLM, that can excel in customizing the course structure, explanations, forms of content, etc on a specific domain like ...
- 1
-4
votes
0
answers
32
views
High Test accuracy value but bad at prediction [closed]
import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.applications import MobileNet
batch_size = 8
num_epochs = 20
num_classes = 4
...
-1
votes
0
answers
28
views
Auto-Encoder Learning Issue – Validation Better Than Training? [closed]
I'm developing an autoencoder that encodes tree-based expressions into a latent vector and decodes them back to their original form.
My dataset consists of genetic programming trees formatted like ...
- 1
-3
votes
0
answers
39
views
Is Cursive Writing Character segmentation even possible using this strip wise OCR (or other Classification model) based technique? [closed]
I need to extract instances of characters from a sample of user's handwritten text. In effect, I am trying to create a database of how a user writes a specific letter from the submitted text sample. ...
-2
votes
0
answers
65
views
how to use goal seek functionality in machine learning model [closed]
I am trying to implement goal seek functionality in my regression model. My model is already built and got the forecasting value for Jan25, Feb25, March25 now i want to do reverse if i know target ...
- 3
1
vote
0
answers
87
views
How to remove dotted boundary boxes while preserving handwritten digits/characters for OCR
I'm working with images that contain handwritten digits and letter 'x' (signifying that that box should be considered 'empty') where each character is written inside a dotted box. I'm trying to remove ...
- 11
-2
votes
0
answers
24
views
AutoKeras need to rewrite Keras code in python
i have trained data with Autokeras and save the model.
import pandas as pd
from numpy import asarray
from pandas import read_csv
from sklearn.model_selection import train_test_split
from sklearn....
-4
votes
0
answers
25
views
How to do behavioural analysis on a smartphone? [closed]
I need to track social media usage patterns, and engagement sentiment of the individual to monitor the person's behaviour and give a mental health score based on the data analysed.
There is one ...
-5
votes
0
answers
47
views
How to interpret MASE and RMSE values for a Machine Learning model [closed]
After fitting the Machine learning models and while evaluating the predictions using the metrics RMSE, MASE, R2 score
The MASE value is low for the same model, but the R2 score is also low, and the ...
- 91
1
vote
0
answers
33
views
How Come Memory Footprint of a Tensorflow Model is Increasing After tflite Conversion?
Trained a simple tensorflow model containing some LSTM and Dense feed-forward layers. After training, I am quantising and converting the model to a tf.lite format for edge deployment. Here is the ...
- 1,632
1
vote
1
answer
30
views
What is the box constraint in the output of a support vector machine in matlab?
In matlab, the function fitcsvm trains a support vector machine.
In output, there is a component BoxConstraints. I have read the post this post
and understood the meaning of box constraint.
But in the ...
- 137
0
votes
0
answers
19
views
DQN agent: loss decreases, cumul. reward stagnates, q-values are very similar over all actions and get higher and higher
My loss is decreasing, but my agent isnt learning in a very easy environment... The cumulative reward stagnates and when printing the q-values they are identical or at least very similar over all ...
- 1
0
votes
0
answers
43
views
How to Convert 'data.ilearner' Model to 'model.pkl' in Azure ML?
I created a pipeline in Azure ML that trains a model using Boosted Decision Tree Regression. From my understanding, the model is saved as data.ilearner.
However, I am unable to convert this model into ...
- 671
-3
votes
0
answers
32
views
Temporal Fusion Transformer: 0.45 R² (Yahoo) vs. Negative R² (Investing.com) [closed]
I'm using a Temporal Fusion Transformer (TFT) to predict daily FX price changes. It works well on Yahoo Finance data (R²=0.45) but fails badly on Investing.com data (negative R²), despite identical ...
- 43
0
votes
0
answers
35
views
Does reducing the number of categories improve YOLO performance? [closed]
Suppose I have set of images that contain two similar categories and one category that is different from the first two. Think cats, dogs and tennis balls. Now I need to train a YOLO object detector ...
- 171
-2
votes
0
answers
30
views
Implementing the Discriminant Algorithm for Reduced Multidimensional Features in Python [closed]
I need to implement the Discriminant Algorithm for Reduced Multidimensional Features proposed in the paper https://www.jstage.jst.go.jp/article/nolta/1/1/1_1_37/_pdf/-char/en as Algorithm 2. I am ...
-1
votes
1
answer
21
views
kNN search on a very large dataset with a torchscript compatible library [closed]
I am trying to run a k-NN search for a very large dataset (1e5 points). Libraries like PyKeOps work fine in terms of memory and time but unfortunately it is not TorchScript compatible. Is there any ...
- 137
0
votes
2
answers
73
views
Pipeline FutureWarning: This Pipeline instance is not fitted yet [closed]
I am working on a fairly simple machine learning problem in the form of a practicum. I am using the following code to preprocess the data:
from preprocess.date_converter import DateConverter
from ...
- 11
-2
votes
0
answers
20
views
Unable to Download and Unzip Datasets (Human3.6M, Penn Action, MPI-INF-3DHP) in Colab [closed]
I’ve been trying to download and unzip the following datasets: Human3.6M, Penn Action, and MPI-INF-3DHP in Google Colab. However, I keep encountering the same error, which I suspect is related to the ...
-2
votes
0
answers
21
views
Anomaly Detection in unsupervised setting [closed]
I have been trying to solve a problem in anomaly detection in unsupervised setting and it seems I have high cardninality variables and I want to fine tune the mistral 7b using the data.
I have this ...
-3
votes
0
answers
22
views
How do we balance the trade-off between AI model transparency and performance in critical applications? [closed]
In recent years, we've witnessed a remarkable surge in the performance of deep learning models, particularly in critical sectors such as healthcare, finance, and autonomous systems. However, this ...
1
vote
0
answers
44
views
Runtime Error while trying to train RTDetrV2 with Transformer
I am trying to train RTDetrV2 for detection on water meter digit. I use an ipynb file form here https://colab.research.google.com/github/roboflow-ai/notebooks/blob/main/notebooks/train-rt-detr-on-...
1
vote
2
answers
53
views
Why does my Keras model show 4 parameters instead of 2?
I'm training a simple Keras model with one Dense layer, but when I call model.summary(), it shows 4 parameters instead of 2.
import tensorflow as tf
import numpy as np
# Generate dummy data
X_train = ...
- 49.7k
0
votes
1
answer
24
views
Why VotingClassifer performance with voting set to "hard" is different with different weights?
I wanted to test VotingClassifier from sklearn and comparae performance with different parameters. I used param grid and then I notice something unintelligible.
I prepared three classifiers
gnb = ...
0
votes
0
answers
34
views
Installing LlaVA
What are the right dependencies and sequence of installation for LlaVA? I have tried multiple options but keep getting the below error when trying to source the model:
Code:
from llava.model.builder ...
- 29
1
vote
0
answers
31
views
Stable torch, ultralytics, tensorflow packages for requirements.txt
I'm writing a requirements.txt for a real-time sports video analysis project and testing various libraries such as YOLOv8, Torch, TensorFlow, etc.
I've tried multiple Python versions but encountered ...
- 41
0
votes
3
answers
31
views
How to let PyTorch learn coefficients variables for the loss outside of nn.Module?
I have a combined loss funcation like this:
loss = alpha * loss0 + beta* loss1 + gamma * loss2 + delta* loss3
I would like to make alpha, beta, gamma, and delta learnable parameters. Notice that ...
- 11.4k
-1
votes
1
answer
42
views
Anomaly detection ML training with none or few anomalies [closed]
When working with anomaly detection of unsupervised non-sequential data using Isolation forest, Local outlier factor with low-variance variables such as temperature and humidity.
And we train the ...
- 1
0
votes
1
answer
19
views
In ordinal encoder what does handle_unknown= use_encoded_values do?
I've done my research about but I'm not satisfied with the answer I looked up both on the documentation and gemini. use_encoded_value what does it mean? Do I have to pass an argument to act as an ...
1
vote
1
answer
23
views
BigQuery ML Time Series model evaluate keeps returning null
I’m using BigQuery ML to train an ARIMA_PLUS model for forecasting CPU usage. The model trains successfully, but when I run ML.EVALUATE, all result values are NULL.
Model Training Query
CREATE OR ...
-1
votes
0
answers
18
views
NN Regression Training loss initial increase [closed]
I'm tuning up a NN to predict growth within a dataset, but I am having two issues in my experimentation; 1) Running the experiments multiple times is yielding loss values varying by as much as 0.5 ...
2
votes
0
answers
96
views
Using binary ({0,1}) weights in a Neural Network layer [closed]
I want to create a custom linear (not fully connected) layer in Pytorch. What is special in this layer is that I need its weights to be binary (0 or 1), so that this layer acts as a "mask", ...
- 109
1
vote
0
answers
18
views
How to prepare irregularly spaced time-series data for classification using LSTM?
I have the variable holding 215 days worth of data structured like this: processed_data is a cell array of size 215×1, holding cells, where each cell contains data for a given day. Each cell (day) has ...
- 61