# Tag Info

### Could you explain if this plot is good or bad. It is a sentiment analysis modelusing LSTM layers

"Good" or "bad" is always relative in data science. You need to establish a benchmark for comparison. First of all, you need to know that accuracy is not a very good performance ...
• 153

### Usage of Word2Vec

Sure, you can average the word2vec vectors of all words in the sentence and train a linear classifier with labeled data. Before doing that, you may remove stopwords that do not add meaning. This ...
• 19k
Accepted

### Removing seasonality in time series forecasting

Removing seasonality is not something you are obliged to do. It really depends on the model. The idea of decomposing time series (you are not actually removing seasonality, it is simply a component ...
• 438
Accepted

### Why so discrepancy between ARIMA and LSTM in time series forecasting?

Arima and LSTM are very different and there could be some tips to improve results. Have you tried relative values instead of raw values? For instance: ...
• 4,299

### How to arrange multiple multivariate time series of different length before passing it to Keras LSTM layer

To use multiple multivariate time series with different lengths and timestamps as input to a Keras LSTM model, you can follow these steps: Pad the time series to the same length: You can pad each ...
• 394

### Could you explain if this plot is good or bad. It is a sentiment analysis modelusing LSTM layers

It is not a good plot because: The axis lack labels. The legend labels are not very meaningful, "Validation" and "Test" would be better choices. The plot relies solely on color to ...
• 131
Accepted

### Early anomaly detection / Failure prediction on time series

This is a standard problem in the field of Predictive Maintenance, and there are several ways to model it. The key question is whether there is predictive information present in the data-stream at all....
• 1,292
Accepted

### Neural network / machine learning approach to model specific sequencing-classification problem in industry

I would try to apply techniques from the "changing point problem" world. In this kind of problem, you try to identify times when the probability distribution of a stochastic process or time ...
• 536
Accepted

### RNN basic doubt

No, they will not have the same final output. Although the weights of the RNN are the same for each time step and the words are the same, their order is not and therefore the inputs and hidden states ...
• 19k

### What are the evaluation metrics we can use for RNN models?

Frame challenge: The metric you use to evaluate your models should be based on the task you're trying to solve and things like distribution of labels in the data, not the type of model. As long as the ...
• 567
Accepted

### does "unravelling" lstm units still mean one unit

The "unraveling" you are referring to is just to illustrate how the different time steps of the input are received and processed. It doesn't have anything to do with the number of units. The ...
• 19k

### Usage of Word2Vec

Word2Vec word vectors were being used for classification purposes in the early days. We could take word vectors of all the words in the sentence to get the resultant vector which could give us some ...

### Validation loss is not decreasing

It may be that you need to feed in more data, as well. If the model overfits, your dataset may be so small that the high capacity of the model makes it easily fit this small dataset, while not ...
• 151

### LSTM model accuracy checking

That is the graph of the loss curves. It clearly shows that the model is overfitting because the test curve is going up. This is bad. It means that your model will perform badly in data that is not ...
• 19k
Accepted

### Should I annotate additional information besides the categories I already need in a text?

It might depend on the algorithm you choose and on how various your data is. Solution 1: If every potential case is precisely identified, it could be better to classify every field precisely. Solution ...
• 4,299
Accepted

### Is it possible to train a RNN using multiple time series?

Yes, you can use a Multivariate RNN. Multivariate RNN In this architecture multiple sequential features (i.e., a number of sequneces) as an input to your recurrent layers. Taking pytorch as a ...
• 1,918

### LSTMs how to forecast out N steps

So this is what I have to predict 96 samples into the future which is 24 hours 15 minutes at a crack: ...
• 381

### Theory behind time Series Test dataset being the last x%

Splitting the test set into two separate chunks in the middle and at the end may not provide any additional benefits and may even introduce some bias, as it may not ...
• 41
1 vote

### RNN/LSTM with multiple targets and varying sequence

From my understanding of your problem description, I believe a similar problem has been tackled by the authors of RAINDROP. You can take a look at that. The caveat is that the authors use a GNN based ...
• 11
1 vote
Accepted

### LSTM Feature engineering: using different Knowledge Graph data types

As general points: Multivariate RNN: You can use multiple sequential features as an input to your recurrent layers. Taking pytorch as a reference, you can see that the input of LSTM object is a ...
• 1,918
1 vote
Accepted

### RNN for continuous, real-time learning without pre-training

I would look into the Markov Chain This is quite a commonly solved problem with many approaches.
• 178
1 vote

### Val loss initially decreases, then increases

The point where the validation loss starts to grow is where the training starts to overfit the model, i.e. it is memorizing your training data and getting worse at generalizing to new data. There are ...
• 138
1 vote
Accepted

### Transformers vs RNN basic doubt

There are multiple concepts mixed in your question. Contextual vs. non-contextual word embeddings: word2vec is a non-contextual approach to obtaining token embeddings. This means that a specific word ...
• 19k
1 vote
Accepted

### Is an LSTM cell autoregressive?

It is autoregressive because its output at time $t - 1$, $h_{t - 1}$, is received as input for the computation at time $t$ and used to generate $h_t$. The fact that there are other inputs like $x_t$ ...
• 19k
1 vote

### Deep learning approach for calibration of raw data using reference measurements and a recurrent neural network (LSTM)

The difference between high-quality sensors and low-quality ones could be regarded as noise. That's why you can either use a noise suppression algorithm or autoencoders based on high-quality sensors ...
• 4,299
1 vote

### I tried loading my saved .h5 model and predicting with that model, i'm getting error list index out of range

I think the error might be caused because you are giving a plain text as input to your model. While preparing a model and training it, I usually use a tokenizer from ...
1 vote
Accepted

### Multiple features in LSTM

In the case where there are multiple features, the LSTM processes each feature independently, with its own set of weights and biases. The LSTM uses the same computational graph for each feature, but ...
• 394
1 vote
Accepted

### In LSTM why h_t output twice?

ht was initially defined as a differential function, which value is the same in output and in the next LSTM cell. LSTM uses the previous steps in a sequential way and chooses whether to memorize or ...
• 4,299
1 vote

### Features and LSTM

Did you normalize your data with a min-max scaler? LSTM is a complex neuron, and its size should be adapted enough to your data: very simple models could under-perform because LSTMs are not suited for ...
• 4,299
1 vote

### LSTM basic doubt

First of all LSTM does not know the word it received, not at least in the sense that you think. LSTM like all neural net cell works with numeric vector representation. Even if you would build a ...
• 4,593

Only top scored, non community-wiki answers of a minimum length are eligible