All Questions
36,629
questions
0
votes
0
answers
2
views
Explanation : Simpler models beat BERT base
I have been trying to train different models for a multi-class classification task of texts. My data set consists of rows of text and its label. The texts are short sentences.
I tried the following ...
- 101
0
votes
0
answers
9
views
LSTM Layer producing same outputs for different sequences
Currently I try to train on a multi-label language task with imbalanced class distribution. I have the following model, where I removed some of the feed forward layers to decrease factors in the chain ...
0
votes
1
answer
6
views
RuntimeError: mat1 and mat2 shapes cannot be multiplied (25x7 and 1x512)
I am dealing with multivariate time series forecasting using Transformers.
below is my code step by step:
After some preprocessing and windowing time series dataset …
1- Creating Mask function
...
- 1
0
votes
1
answer
10
views
Images used in training CNN model
I am training a CNN with RGB images, however, when i plot them, they display in a bluish color. How can I have these display in RGB. Also do you think this affects the model accuracy?
The code i am ...
- 1
1
vote
1
answer
14
views
How to represent output layer if the action size changes dynamically?
I am new to ML and I need to train a chess agent using proximal policy optimization. Board is represented as string and the environment gives a list of valid moves for each step.
...
- 11
0
votes
0
answers
6
views
Understanding processors in huggingface tokenizer library
tl;dr
What are the :0 and :1 in the following huggingface processors reference usage given on their page:
...
- 821
0
votes
0
answers
14
views
Predict best chess move using RNNs
I am trying to do a project with AI: in which during any certain moment of a chess game i can predict, using a RNN trained on a kaggle dataset, the best possible move that i can make.
I am having ...
0
votes
0
answers
7
views
Model performance impact on social discrimination?
I am currently working on a project where the data concerns people and the dataset contain personnal data with sensitive attributes. (typically: age, sex, handicap, race).
Now it seems there are ...
- 2,576
0
votes
0
answers
3
views
Custom loss function for multi label classification in catboost?
I have a data frame which I want to use for multi class classification problem. There are total 6 classes (say a, b, c, d, e, f). I want to improve the precision for three classes (say a, b, c) i.e. ...
- 101
0
votes
0
answers
6
views
How can I leverage machine learning for log analysis?
I am new to data science and trying to find possibilities of using datascience in tasks. I have a set of logs which I want to convert to json. The logs are more or less of same format and I can write ...
- 101
0
votes
1
answer
17
views
Implementing Dropout in Keras
I think I am not conceptually understanding "Dropout" in neural networks. I was under the assumption that a keep rate of 0.8 would set 20% of all the neurons to 0 for each training example.
...
0
votes
0
answers
12
views
Should I log transform the data if the dataset is small?
Suppose I have a dataset with only about 25 rows.
All the relevant columns are all integers. The ranges of them go from 1 million to 100 million, and the distributions are all skewed right.
In this ...
- 1
0
votes
0
answers
23
views
Improving performance of anomaly detection using dataset?
I am leveraging an isolation forest model from the scikit-learn library for anomaly detection in a time series dataset where each point in the dataset is a data frame. However, I possess additional ...
- 101
0
votes
0
answers
5
views
How are the words defined in the sentencepiece algorithm?
I am not able to understand how the sentencepiece algorithm solves the problem of handling the languages without a clear-cut concept of words
My exact confusion is:
It is mentioned that one of the ...
- 821
0
votes
0
answers
17
views
Purpose of Azure Cognitive Search
Azure provides a service called Cognitive Search which is an intelligent AI-based search service based on advanced NLP.
I tried this feature. And to make the search as efficient as possible, it ...
- 275
0
votes
1
answer
21
views
VGG16 Transfer Learning for image binary classification - suspected overfitting
I'm using VGG16 for transfer learning on a binary image classification task about human posture. The sample totaled about 2,000 images, with about 900 and 1,000 images in each category, respectively. ...
- 1
0
votes
0
answers
16
views
Working of 'Interpolation' to impute missing values & the importance of the order of data points for the process
Definition of interpolation -
Interpolation predicts values at a point by studying its neighbouring points (within the same column), as opposed to data modeling where all the columns are taken into ...
- 275
0
votes
0
answers
5
views
R, Caret Predict(), and errors
So I've been playing around with/testing out a few ideas. I just wasn't sure though if there is a simple command you can use to have caret like send you back a list (or data.frame) of all the errors ...
- 11
0
votes
0
answers
6
views
How can I detect patterns in a csv of numeric values with a success failure indicator?
Hello everyone I have a need and I would like help to identify the technological approach, I tentatively consider that I need unsupervised ml for the following: I have a data dataset about investment ...
0
votes
1
answer
22
views
Why the training accuracy stays high but validation accuracy does not change?
I have a binary classification problem. I get ROI mammogram images and then apply a decomposition algorithm and as output I get 5 images which summation of them results in the original image. Now, ...
- 1
1
vote
0
answers
10
views
How do I achieve MAE < 8 using ResNet50 on this dataset?
I've been up all night trying to achieve MAE < 8 for this age recognition dataset: https://people.ee.ethz.ch/~timofter/publications/Agustsson-FG-2017.pdf
It is for an online class I am taking, the ...
- 111
0
votes
1
answer
11
views
CLIP Paper or CLAP Paper (I dont understand the loss function) - Can you help?
Can somebody help me understand this Contrastive Learning pretraining paper? This explanation comes from https://arxiv.org/pdf/2206.04769.pdf (page 2).
I understand that they apply an audio encoder ...
- 65
0
votes
0
answers
17
views
why my training and validation loss curve looks like lognormal distribution?
I trained an XGBoost model and my training and validation curve looks like this? Is something weird I am doing? I have always seen it going from high to low or like a U-shape incase of overfitting.
- 101
0
votes
0
answers
31
views
My machine learning model cannot classify data that it has never seen before
I am creating a spam identifier to detect wheter an email is malicous. The issue I have is my model using the RandomForest Classifier showed it was 99% accurate.
csv: https://www.kaggle.com/datasets/...
2
votes
1
answer
108
views
Higher level sentence similarity (meaning instead of 'just' embeddings)
I am looking for the correct model / approach for the task of checking if two sentences have the same meaning
I know I can use embeddings to check similarity, but that is not what I am after. I ...
- 123
0
votes
0
answers
10
views
Inverse Laplacian of xarray
my question is related to calculating the inverse Laplacian of a xarray. Specifically, I want to solve the equation ∇^2 s = h+r+e for s in Python (where ∇^2 denotes the laplacian). Here, s is an ...
0
votes
1
answer
19
views
Use prediction after using get_dummies in pandas?
I found similar question on this topic but no answer was helpful.
I had a data frame with a categorical column in it with 5 different values. I used get_dummies and used linear regression for ...
- 71
1
vote
1
answer
15
views
how to get rid of unseen labels
I am trying to perform naive bayes, but my testing set has a large amount of unseen labels. I thought the easiest way fix this issue is to remove them from the dataset but I get the error, TypeError: ...
- 11
0
votes
0
answers
26
views
Best model for regression in this case?
I am doing some modeling to predict a variable of interest given a big set of features (500) for which I expect a considerable amount of interactions happening at least among some of them.
I first ...
- 11
1
vote
1
answer
36
views
How to filter a dataframe and get a sub data frame based on frequency?
I have a data-frame. Lets call it df. My data frame has a column with categorical values e.g. a column named Place. But the problem is that there are many different (string) values in that column and ...
- 71
0
votes
1
answer
18
views
Question about bias-variance trade off
This question is from a uni module about machine learning. I'm a bit stuck as I can't relate it to the bias-variance trade-off, to me the question implies all models have something to do with the ...
0
votes
0
answers
9
views
Could we use another attention module on the outputs of the attention heads in the transformer architecture?
Before concatenating the heads in MHAtt, could we add another attention module with the heads as input to combine them?
Thus, for each head we would get a value matrice enriched with the outputs of ...
- 1
0
votes
0
answers
8
views
How to make duplicates unique in a DataFrame robustly?
I have the following table.
I was able to parse it and display it as a DataFrame with pandas in python
I need to change the Ref for every value that is duplicate such that it becomes unique.
For ...
- 101
0
votes
0
answers
22
views
Algorithms from R-statistics package Caret R Package- LVQ algorithm, is there similar in Python
In the R-statistics package : Caret R Package, they have the LVQ algorithm that is used
for the purpose of "Feature Selection".
I have used this to do some data science in R-stats over 6 ...
- 103
0
votes
1
answer
17
views
Is it possible to use the SODA tool as open source?
I was reading about the SODA tool and was wondering if it was free to use. As far as I understand, it has a SODA library and a SODA core, but they are all just clients for connecting to a cloud server ...
- 1
0
votes
0
answers
11
views
Is there an appropriate hypothesis test: Two samples, one only with one datapoint, non-normal?
I have two 'samples'.
The first consists of approx. 400 physical measurements of a quantity (taken over one hour, and the real situation is not a steady state). They show a very skewed distribution ...
- 1
0
votes
1
answer
32
views
Is improving a Neural Network really just "trial and error"?
After asking on StackOverflow, I was redirected here, so I'm reposting this question.
I am a PhD student in Computational Physics and I've started to study a bit of Neural Networks, and decided to try ...
- 101
0
votes
0
answers
7
views
Demand Forecasting methods for hundreds of time series
I have TFL cycling dataset for the time period from JAN 2023 to JUNE 2023. I would like to forecast the demand or expected no. of trips for each station at every hour of the day.
Post some data ...
- 197
0
votes
1
answer
18
views
Working with functions
What are some good sources for learning how to deal with functions in Python and R. I have used them a lot of times and they can be a bit challenging. I like to write out a plan of what I want done ...
0
votes
1
answer
23
views
Is there a derichlet-tree learning algorithm written in Python
I want to learn a decision-tree based on dirichlet distribution (namely a derichlet-tree). Which Python (or possibly other) libraries/packages enables such algorithm?
- 17
0
votes
1
answer
22
views
CNN training accuracy flatlines
I'm training a CNN from scratch to do tagging of images. And my training is going nowhere. I was hoping someone could help me identify an obvious error.
I would like to end up with a network that ...
- 101
0
votes
0
answers
8
views
How to control and optimize optuna
i'm here with a pretty open question.
I'm using Optuna to fine tune a Catboost Regressor and i've found, trying a set of parameters by hand, that the "best params" it outputs are not the ...
- 101
0
votes
1
answer
33
views
Optimizing YOLO for Diverse Takeaway Item Detection: Single vs Type-Based Classes
I'm using YOLO to detect various 'takeaway' items, currently all marked as class 0, which vary very widely(like groceries) in shape and type. Considering the introduction of new items on which the ...
0
votes
1
answer
21
views
In lightgbm,what is a uniform drop?
I read through document of lightgbm,it just tells this parameter, but didn't give much explanation for it.
The explanation:
...
- 393
0
votes
1
answer
23
views
Are my regression metrics value correct?
So im using a dataset for Wine Prediction where im using Linear Regression model to predict the prices.
These are the steps i'm using:
...
0
votes
0
answers
22
views
Cross validation and standard error
I'm performing a CV and I need to report the score with its error, but it only returns the score's average from the folds. Is there a way to either show each folder's score or find the standard error ...
0
votes
1
answer
18
views
Does learning rate depend on input and output range?
I watched hours of videos on gradient descent and still feel pretty confused. Let's say I have a "model":
y = x * w
I use 2 as my target ...
- 103
1
vote
1
answer
31
views
Minimizing error in cosine similarity
Presume I have a vector space, and I am attempting to compress it into a latent vector space, while minimizing error in cosine similarity between entries. Suppose that I know the actual cosine ...
- 13
0
votes
1
answer
33
views
Python SK-Learn KNN Imputer ( "ValueError: could not convert string to float: )
I have data with missing values.
All columns are integer, except for a column that has missing values.
These missing values, were set with a "?" which was converted to NaN using the Numpy ...
- 103
0
votes
1
answer
15
views
Number of feature extraction layers in CNN
In a course I took about machine learning, we normally used about 2 feature extraction layers for image classification tasks, using MNIST or ...
- 101