Select your cookie preferences

We use essential cookies and similar tools that are necessary to provide our site and services. We use performance cookies to collect anonymous statistics, so we can understand how customers use our site and make improvements. Essential cookies cannot be deactivated, but you can choose “Customize” or “Decline” to decline performance cookies.

If you agree, AWS and approved third parties will also use cookies to provide useful site features, remember your preferences, and display relevant content, including relevant advertising. To accept or decline all non-essential cookies, choose “Accept” or “Decline.” To make more detailed choices, choose “Customize.”
Sign in
Sign in
or
Create a new account
About
Categories
Delivery Methods
Solutions
Resources
 Your Saved List Become a Channel Partner Sell in AWS Marketplace Amazon Web Services Home Help

Amazon Sagemaker

Amazon SageMaker is a fully-managed platform that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale. With Amazon SageMaker, all the barriers and complexity that typically slow down developers who want to use machine learning are removed. The service includes models that can be used together or independently to build, train, and deploy your machine learning models.

product logo

Time Series Forecasting (LNN) Free trial

By:
fg-research
Latest Version:
1.7
Perform time series forecasting in SageMaker with Liquid Neural Networks.
Continue to Subscribe

    Product Overview

    This algorithm performs time series forecasting with Liquid Neural Networks (LNNs). It implements both training and inference from CSV data and supports both CPU and GPU instances. The training and inference Docker images were built by extending the PyTorch 2.1.0 Python 3.10 SageMaker containers. The Docker images include software licensed under the Apache License 2.0. The algorithm can be used for both univariate and multivariate time series and supports the inclusion of external features.

    Key Data

    Version
    1.7
    Usage Instructions
    Release Notes
    By
    fg-research
    Categories
    ML Solutions
    Time-series Forecasting
    Type
    Algorithm
    Fulfillment Methods
    Amazon SageMaker

    Highlights

    • The algorithm performs time series forecasting with LNNs directly from CSV data.

    • The algorithm allows tuning the model hyperparameters to optimize performance on custom datasets.

    • The algorithm supports CPU, GPU and multi-GPU training.

    Not quite sure what you’re looking for? AWS Marketplace can help you find the right solution for your use case. Contact us

    Pricing Information

    Use this tool to estimate the software and infrastructure costs based your configuration choices. Your usage and costs might be different from this estimate. They will be reflected on your monthly AWS billing reports.

    Contact us to request contract pricing for this product.


    Estimating your costs

    Choose your region and launch option to see the pricing details. Then, modify the estimated price by choosing different instance types.

    Version
    Region

    Software Pricing

    Algorithm Training$4.99/hr

    running on ml.m5.2xlarge

    Model Realtime Inference$0.99/hr

    running on ml.m5.2xlarge

    Model Batch Transform$4.99/hr

    running on ml.m5.2xlarge

    Infrastructure Pricing

    With Amazon SageMaker, you pay only for what you use. Training and inference is billed by the second, with no minimum fees and no upfront commitments. Pricing within Amazon SageMaker is broken down by on-demand ML instances, ML storage, and fees for data processing in notebooks and inference instances.
    Learn more about SageMaker pricing

    SageMaker Algorithm Training$0.461/host/hr

    running on ml.m5.2xlarge

    SageMaker Realtime Inference$0.461/host/hr

    running on ml.m5.2xlarge

    SageMaker Batch Transform$0.461/host/hr

    running on ml.m5.2xlarge

    About Free trial

    Try this product for 5 days. There will be no software charges, but AWS infrastructure charges still apply. Free Trials will automatically convert to a paid subscription upon expiration.

    Algorithm Training

    For algorithm training in Amazon SageMaker, the software is priced based on hourly pricing that can vary by instance type. Additional infrastructure cost, taxes or fees may apply.
    InstanceType
    Algorithm/hr
    ml.m4.4xlarge
    $4.99
    ml.c5n.18xlarge
    $4.99
    ml.g4dn.4xlarge
    $4.99
    ml.m5.4xlarge
    $4.99
    ml.m4.16xlarge
    $4.99
    ml.m5.2xlarge
    Vendor Recommended
    $4.99
    ml.p3.16xlarge
    $4.99
    ml.g5.xlarge
    $4.99
    ml.g5.12xlarge
    $4.99
    ml.g4dn.2xlarge
    $4.99
    ml.g5.4xlarge
    $4.99
    ml.c5n.xlarge
    $4.99
    ml.m4.2xlarge
    $4.99
    ml.c5.2xlarge
    $4.99
    ml.p3.2xlarge
    $4.99
    ml.c4.2xlarge
    $4.99
    ml.g4dn.12xlarge
    $4.99
    ml.p4d.24xlarge
    $4.99
    ml.m4.10xlarge
    $4.99
    ml.c4.xlarge
    $4.99
    ml.m5.24xlarge
    $4.99
    ml.c5.xlarge
    $4.99
    ml.g4dn.xlarge
    $4.99
    ml.g5.48xlarge
    $4.99
    ml.p2.xlarge
    $4.99
    ml.m5.12xlarge
    $4.99
    ml.g4dn.16xlarge
    $4.99
    ml.p2.16xlarge
    $4.99
    ml.c4.4xlarge
    $4.99
    ml.g5.8xlarge
    $4.99
    ml.m5.xlarge
    $4.99
    ml.c5.9xlarge
    $4.99
    ml.g5.16xlarge
    $4.99
    ml.m4.xlarge
    $4.99
    ml.c5.4xlarge
    $4.99
    ml.p3.8xlarge
    $4.99
    ml.m5.large
    $4.99
    ml.c4.8xlarge
    $4.99
    ml.c5n.2xlarge
    $4.99
    ml.p2.8xlarge
    $4.99
    ml.g4dn.8xlarge
    $4.99
    ml.c5n.9xlarge
    $4.99
    ml.c5.18xlarge
    $4.99
    ml.g5.2xlarge
    $4.99
    ml.c5n.4xlarge
    $4.99
    ml.g5.24xlarge
    $4.99

    Usage Information

    Training

    The input dataset should be provided as a CSV file. Each column represents a time series, while each row represents a time step. All the time series should have the same length and should not contain missing values. The column headers should be formatted as follows:

    • The column names of the (mandatory) target time series should start with "y" (e.g. "y1", "y2", ...).
    • The column names of the (optional) feature time series should start with "x" (e.g. "x1", "x2", ...).
    • The column containing the time spans (optional) should be named "ts".

    Metrics

    Name
    Regex
    train_mse
    train:mse ([0-9\\.]+)
    train_mae
    train:mae ([0-9\\.]+)
    valid_mse
    valid:mse ([0-9\\.]+)
    valid_mae
    valid:mae ([0-9\\.]+)

    Channel specification

    Fields marked with * are required

    training

    *
    Input modes: File
    Content types: text/csv
    Compression types: None

    validation

    Input modes: File
    Content types: text/csv
    Compression types: None

    model

    Input modes: File
    Content types: "model.tar.gz" file from a previous training job
    Compression types: None

    Hyperparameters

    Fields marked with * are required

    context-length

    *
    The length of the input sequences
    Type: Integer
    Tunable: Yes

    prediction-length

    *
    The length of the output sequences
    Type: Integer
    Tunable: Yes

    sequence-stride

    *
    The period between consecutive output sequences
    Type: Integer
    Tunable: Yes

    backbone-layers

    *
    The number of hidden layers of the backbone neural network
    Type: Integer
    Tunable: Yes

    backbone-units

    *
    The number of hidden units of the backbone neural network
    Type: Integer
    Tunable: Yes

    backbone-activation

    *
    The activation function of the backbone neural network
    Type: Categorical
    Tunable: Yes

    backbone-dropout

    *
    The dropout rate of the backbone neural network
    Type: Continuous
    Tunable: Yes

    hidden-size

    *
    The number of hidden units of the neural network heads
    Type: Integer
    Tunable: Yes

    minimal

    *
    If set to 1, the model will use the CfC direct solution
    Type: Integer
    Tunable: Yes

    no-gate

    *
    If set to 1, the model will use a CfC without the (1 - sigmoid) part
    Type: Integer
    Tunable: Yes

    use-ltc

    *
    If set to 1, the model will use an LTC instead of a CfC
    Type: Integer
    Tunable: Yes

    use-mixed

    *
    If set to 1, the model will mix the CfC RNN-state with an LSTM
    Type: Integer
    Tunable: Yes

    lr

    *
    The learning rate used for training
    Type: Continuous
    Tunable: Yes

    lr-decay

    *
    The decay factor applied to the learning rate
    Type: Continuous
    Tunable: Yes

    batch-size

    *
    The batch size used for training
    Type: Integer
    Tunable: Yes

    epochs

    *
    The number of training epochs
    Type: Integer
    Tunable: Yes

    Model input and output details

    Input

    Summary

    The inference algorithm takes as input a CSV file containing the time series. The CSV file should have the same format as the one used for training.

    Input MIME type
    text/csv
    Sample input data
    Expand all input descriptions
    Field name
    Target time series
    Description
    The time series to be predicted.
    Required
    Yes
    Data type
    Continuous (Float)
    Range
    Field name
    Feature time series
    Description
    The time series of the external features.
    Required
    No
    Data type
    Continuous (Float)
    Range
    Default value
    If the features are not provided, the algorithm will only use the past values of the target time series as input.
    Field name
    Time spans
    Description
    The time intervals between consecutive time steps.
    Required
    No
    Data type
    Continuous (Float)
    Range
    Default value
    If the time spans are not provided, the algorithm will assume that the observations are equally spaced.

    Output

    Summary

    The inference algorithm outputs the predicted values of the time series and the standard deviation of the predictions. The outputs include the out-of-sample forecasts beyond the last time step of the inputs.

    Output MIME type
    text/csv
    Sample output data
    Expand all output descriptions
    Field name
    Mean
    Description
    Predicted values of target time series.
    Always returned
    Yes
    Data type
    Continuous (Float)
    Range
    Field name
    Standard Deviation
    Description
    Standard deviation of the predictions.
    Always returned
    Yes
    Data type
    Continuous (Float)
    Range

    Sample notebook

    Additional Resources

    End User License Agreement

    By subscribing to this product you agree to terms and conditions outlined in the product End user License Agreement (EULA)

    Support Information

    Time Series Forecasting (LNN)

    AWS Infrastructure

    AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.

    Learn More

    Refund Policy

    Subscriptions are not refundable.

    Customer Reviews

    There are currently no reviews for this product.
    View all