Questions tagged [neural-network]
Artificial neural networks (ANN), are composed of 'neurons' - programming constructs that mimic the properties of biological neurons. A set of weighted connections between the neurons allows information to propagate through the network to solve artificial intelligence problems without the network designer having had a model of a real system.
4,346 questions
2votes
0answers
29views
How does fine tuning actually work?
So i’m currently fine tuning a pretrained model with 35k images across 5 classes. Very high class imbalance with one being 73% across the distribution. Handled this with by using a weighted loss ...
- 21
-2votes
0answers
26views
Improve the loss in a neural network model
I've built a neural network model for predicting groundwater quality for certain select contaminants, am attaching the graph of predicted vs actual contaminant values for EC. What are some techniques ...
2votes
0answers
21views
Do i.i.d. assumptions extend to datasets of independently generated sequences in modern sequence models (e.g., RNNs)?
In standard machine learning settings with cross-sectional data, it's common to assume that data points are independently and identically distributed (i.i.d.) from some fixed data-generating process (...
- 101
5votes
1answer
60views
Choosing NN architecture for route correction
Can you help me choose a NN architecture that solves this kind of problem: A route, meaning a sequence of coordinates (e.g. [(34.1, 52.3),(34.101, 52.301), ... ]), ...
4votes
1answer
40views
Can Bounded Activation Functions Like Sigmoid Prevent Exploding Predictions
Just asking here because I'm quite in doubt with my intuition about neural networks's behavior when tuning some parameters. I have been playing with it for quite a while now (around a year), so I ...
- 51
11votes
1answer
800views
Neural network to find errors in training data
My data set consists of an output variable which is categorical with 4 different values and the input variables of which there are roughly 100 and they are boolean, ie True/False. The data set has ...
- 608
2votes
0answers
18views
Multinodal vs modular neural network
I’ve made a couple of neural networks that predict separate target features form the same dataset. All these networks have different hyperparameters and different function transforms for example (log) ...
2votes
0answers
27views
My Neural Network predicts well when there are only no more than two outputs, but for no more
I am using a neural network to predict the parameters (output) of some model given some summary statistics of data (inputs) generated from the model. The model has three parameters. If I hold any one ...
4votes
3answers
145views
Time series predictions with LSTM
I have collection of TEC data.My data sample for example the day1,day2,day3,day4. Case1: I have the following task to do: Training by the consecutive 3 days to predict the each 4th day. Each day data ...
- 105
9votes
4answers
2kviews
Does training a neural network on a combined dataset outperform sequential training on individual datasets?
I have a neural network with a fixed architecture (let's call it Architecture A). I also have two datasets, Dataset 1 and Dataset 2, both of which are independently and identically distributed (i.i.d.)...
0votes
1answer
17views
Best method to analyse user sequence
I have a sequence dataset as the following. These sequences are statuses got approved by clients and they are ordered by date/time. A client can get multiple statuses and jump back to the same status ...
- 31
2votes
0answers
35views
Reducing loss of multiclass classication problem
I have been reading a paper in which they theoretically showed existence of a Neural network model that can perform a algorithm which involves selection of stencils with 100% accuracy. So they ...
- 31
0votes
1answer
31views
Sparse Categorical accuracy shown incorrectly
I am trying to replicate a multiclass classification problem of a paper I am reading. So they provided with the exact matrices and bias vector values and have proved in the paper why there will 100% ...
- 31
0votes
0answers
16views
I would like to build an open source Traffic Signs Dataset solely for research purposes
I've been interested lately in doing research about different neural networks and how to contribute to Autonomous Vehicles, I used a couple of images to train a model and the results were different ...
- 1
0votes
0answers
13views
What's the best model for hourly energy consumption?
I have a dataset of hourly energy consumption from a building from 2010 to 2015. With normal NN I am reaching ~94% accuracy. Didn't try yet with other types of models. Should I try CNN? RNN? ...
