search

zhongkaifu / RNNSharp

RNNSharp is a toolkit of deep recurrent neural network which is widely used for many different kinds of tasks, such as sequence labeling, sequence-to-sequence and so on. It's written by C# language and based on .NET framework 4.6 or above versions. RNNSharp supports many different types of networks, such as forward and bi-directional network, sequence-to-sequence network, and different types of layers, such as LSTM, Softmax, sampled Softmax and others.
BSD 3-Clause "New" or "Revised" License
285 stars 91 forks source link

Older version? #25

Closed lbarwick closed 7 years ago

lbarwick commented 8 years ago

Hi,

Do you have a version available for VS2010 and .NET 4.0?

zhongkaifu commented 8 years ago

Sorry that RNNSharp doesn't support .NET 4.0, since some features, such as SIMD, are not available in the version lower than .NET 4.6.

If you want to use RNNSharp on .NET 4.0, you need to remove those features required higher .NET version, or re-implement them by your owned code.

lbarwick commented 8 years ago

I there a particular ".cs" that I could change for just LSTM? Or is the project all inter-linked?

Lawrence Date: Thu, 14 Jul 2016 06:31:25 -0700 From: notifications@github.com To: RNNSharp@noreply.github.com CC: barwickhome@hotmail.com; author@noreply.github.com Subject: Re: [zhongkaifu/RNNSharp] Older version? (#25)

Sorry that RNNSharp doesn't support .NET 4.0, since some features, such as SIMD, are not available in the version lower than .NET 4.6.

If you want to use RNNSharp on .NET 4.0, you need to remove those features required higher .NET version, or re-implement them by your owned code.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

zhongkaifu commented 8 years ago

You need to change all files if necessary, otherwise, build will fail. If you really want to use it in .NET 4.0, there are two things you need to update:

1. Update "Parallel.For" to "for" statement and "Parallel.Foreach" to "foreach"

2. Replace SIMD code with "for/foreach"

Both of above two things are related to encode/decode speeding up.

lbarwick commented 8 years ago

Hi,

Have you ever tried to apply LSTM to stock prediction?

I am looking into this, but am wondering how to Sequence the input data.

The basic data element I have is the stock price. SO I have created an array of log(Stock Price changes) which I call "Returns".

Do you think I should feed the LSTM with Input of "Last Return" and target output of "Next Return". i.e a single input value and a single target:

Returns = { 0.1, 0.2, -0.1, 0.5, -0.2 ... } Samples = { {Input,TargetOupt}...} Samples = { {{0.1}, 0.2}, {{0.2},-0.1},{{-0.1},0.5}...}

Or provide a longer sequence in the input: Samples = { { {0.1, 0.2, -0.1}, 0.5},{{0.2, -0.1, 0.5},-0.2}}

Any thoughts greatly appreciated,

Lawrence Date: Thu, 14 Jul 2016 06:31:25 -0700 From: notifications@github.com To: RNNSharp@noreply.github.com CC: barwickhome@hotmail.com; author@noreply.github.com Subject: Re: [zhongkaifu/RNNSharp] Older version? (#25)

Sorry that RNNSharp doesn't support .NET 4.0, since some features, such as SIMD, are not available in the version lower than .NET 4.6.

If you want to use RNNSharp on .NET 4.0, you need to remove those features required higher .NET version, or re-implement them by your owned code.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

zhongkaifu commented 7 years ago

I don't think it's good for stock price prediction, but you could try to predict behaviors, such as "Buy", "Hold" and "Sell".