In summary, I have proposed a method to decipher the text of the Voynich manuscript using siamese neural networks. I have created a database of characters and plan to train a model using this database in the coming weeks. This approach has several significant challenges, including collecting and processing data, the unique language and unknown characters of the Voynich manuscript, and the variability of images in the manuscript. However, I believe that this method has potential and I am optimistic about the results that can be achieved. Sharing the results with the community will be crucial for improving the model and contributing to ongoing research on the Voynich manuscript.
Tonight, I am starting the training of my siamese neural network model, with the aim of analyzing the similarities between the characters of the mysterious Voynich Manuscript and those of other languages. This is a crucial step in my quest to decipher this enigmatic and ancient text.
![[Image: Capture-d-cran-2023-08-30-184754.png]](https://i.ibb.co/7KpzGfx/Capture-d-cran-2023-08-30-184754.png)
What makes you so confident that the Voynich was written in the Central Thai language?
(31-08-2023, 01:24 AM)Mark Knowles Wrote: You are not allowed to view links. Register or Login to view.What makes you so confident that the Voynich was written in the Central Thai language?
You are not allowed to view links.
Register or
Login to view.
Not Siamese language
Is 302/302 the number of samples in the training set?
Could you please show some of them, so that we have a chance to understand what you are doing?
Which manuscripts are you using? How did you extract character boxes and label them?
I am also curious about this.
Will you use only (selected) text in the Voynich MS that is written horizontally?
I see there have been some questions about what 'Epoch 1/10 1656/3125' means during the training of a neural network. I will try to clarify this here.
In the context of neural network training, an 'epoch' corresponds to a complete pass through the training dataset. For example, if you have 1000 images in your training dataset, one epoch would correspond to the network passing over all of these 1000 images.
In the example 'Epoch 1/10 1656/3125', '1/10' means that the network is currently in the first epoch of training, and there will be a total of 10 epochs. So, the training dataset will be passed through the network a total of 10 times.
'1656/3125' indicates the progress of the training during this epoch. '1656' is the current batch number and '3125' is the total number of batches. The batch size is the number of training examples passed through the network at once. For example, if you have a batch size of 32, then 32 training examples would be passed through the network each time. So, if you have 100000 training examples, they would be divided into 3125 batches of 32 examples each (100000 / 32 = 3125).
In summary, 'Epoch 1/10 1656/3125' means that the network is currently being trained on the 1656th batch of the first epoch. There will be a total of 3125 batches in this epoch and a total of 10 epochs during the training.
I hope this clarifies things! If you have any other questions, please feel free to ask.
Best regards,
The image where you see character written are different characters of a language that are separated into several different folders with the language used for training to then identify similarities with the voynich characters also trained.
![[Image: Capture-d-cran-2023-08-31-073921.png]](https://i.ibb.co/6rKNWKj/Capture-d-cran-2023-08-31-073921.png)
![[Image: Capture-d-cran-2023-08-31-074028.png]](https://i.ibb.co/g3dngdz/Capture-d-cran-2023-08-31-074028.png)
(31-08-2023, 05:12 AM)Ranceps Wrote: You are not allowed to view links. Register or Login to view. (31-08-2023, 01:24 AM)Mark Knowles Wrote: You are not allowed to view links. Register or Login to view.What makes you so confident that the Voynich was written in the Central Thai language?
You are not allowed to view links. Register or Login to view.
Not Siamese language
Ah. Conjoined twins. I didn't think it was considered politically correct to use the word "Siamese" in that context.
What exactly are you comparing it against?
(31-08-2023, 03:07 PM)davidjackson Wrote: You are not allowed to view links. Register or Login to view.What exactly are you comparing it against?
In fact, what it will do is first take a Voynich character and then try to find a similarity with another character from another language and tell me if it has a similarity with it.