Machine Learning Platform For AI provides end-to-end machine learning services, including data processing, feature engineering, model training, model prediction, and model evaluation. Machine Learning Platform For AI combines all of these services to make AI more accessible than ever.
Machine Learning Platform For AI provides a visualized web interface allowing you to create experiments by dragging and dropping different components to the canvas. The machine learning modeling is a simple, step-by-step procedure, improving efficiencies and reducing costs when creating an experiment.
Using machine learning servitization, Machine Learning Platform For AI allows you to create a complete workflow for enterprise-level machine learning data modeling and application.
Machine Learning Platform For AI provides more than one hundred algorithm components, covering such scenarios as regression, classification, clustering, text analysis, finance, and time series. All of these components are tested on Alibaba Group internal services and provide high stability and high performance.
Powerful Computing Capability
The infrastructure of Machine Learning Platform For AI relies on Alibaba Cloud distributed computing clusters. This allows Machine Learning Platform For AI to handle a large number of concurrent algorithm computing tasks.
Machine Learning Platform For AI includes the following modules:
Machine Learning Platform For AI provides data preprocessing components and feature engineering components for data processing. Data preprocessing components include normalization, standardization, data sampling, and data filtering. Feature engineering components include feature conversion, feature generation, and feature importance evaluation.
Data Mining and Analysis
Machine Learning Platform For AI provides statistics components, machine learning components, and network analysis components for data analysis. Statistics components include statistical analysis and data visualization analysis. Machine learning components include regression, classification, and clustering. Network analysis components include label propagation and largest connected subgraph.
NLP (Natural Language Processing)
Machine Learning Platform For AI provides text processing components for NLP, including word splitting, deprecated word filtering, LDA, TF-IDF, and text summarization.
The architecture of Machine Learning Platform For AI is composed of multiple layers. From bottom to top, the layers include computing engine, distributed computing architecture, algorithm component , and service application. The computing engine layer of Machine Learning Platform For AI relies on the Apsara distributed computing system. This allows Machine Learning Platform For AI to concurrently compute EB-level data.
The distributed architecture computing layer of Machine Learning Platform For AI provides support for multiple distributed computing architectures, such as MPI, MR, and GRAPH. The algorithm component layer provides support for more than one hundred machine learning algorithms.
By using its own algorithms, Machine Learning Platform For AI can support multiple service scenarios such as product recommendation, financial risk management, and advertising at the service application layer.
Machine Learning Platform For AI is billed on a Pay-As-You-Go basis. The total service fee is equal to the billing fee of the component you use multiplied by the number of computing hours.
The computing hours are measured by using the formula Max(vCPU cores, memory size/4) x running time (hours).
Logistic Regression Component Example: Requires 8 vCPU cores and 30GB memory and the component has been running for three hours. The logistic component belongs to the data analysis category, where each component is priced at USD 0.21 per computing hour. The total fee is calculated as USD $5.04 (0.21 x Max(8,30/4) x 3). If this task requires 36 GB memory and 8 vCPU cores, the total fee is USD $5.67 (0.21 x Max(8,36/4) x 3).
Machine Learning Platform For AI Component Billing Table
|Region||Data Processing (USD/ComputeHour)||Data Mining and Analysis (USD/ComputeHour)||NLP (USD/ComputeHour)||Deep Learning (M40)||Deep Learning (P100)|
|China North 2 (Beijing)||0.16||0.21||0.27||-||3.00|
|China East 2 (Shanghai)||0.16||0.21||0.27||2.40||-|
|China South 1 (Shenzhen)||0.16||0.21||0.27||-||-|
|Asia Pacific SE 1 (Singapore)||0.16||0.21||0.27||-||-|
|Asia Pacific SE 3 (Kuala Lumpur)||0.16||0.21||0.27||-||-|
|Asia Pacific SE 5 (Jakarta)||0.16||0.21||0.27||-||-|
A typical user scenario for machine learning is product recommendation. In this scenario, you can use the data preprocessing and feature engineering components of Machine Learning Platform For AI to extract features from customers' historical shopping behavior. This enables you to discover the features that influence such shopping behavior. The machine learning algorithm then verifies whether the features of a customer's behavior on a product match the extracted features. If the features match, then the behavior is determined to be common shopping behavior. Based on the results, Machine Learning Platform For AI recommends the relative products to the customer to increase product sales.
Financial Risk Management
Machine Learning Platform For AI allows you to use the financials algorithm to make an assessment of loan risks. Machine Learning Platform For AI provides the scorecard component for you to calculate the capability of your clients to settle their credit card debt and provides risk indexes to help financial institutions manage risks effectively.
Document classification is a classic text processing scenario in the news industry. Traditionally, all documents have been manually classified, which is inefficient and demands considerable manpower. Machine Learning Platform For AI provides a large number of text analysis components and by learning the existing document classification data, these components can automatically classify documents in a short period of time.
Machine learning Documentation
More information about Machine Learning Platform For AI, go to Document center.
1. How to upload data?
To upload data from the Machine Learning Platform For AI web interface, make sure that the data is less than 20 MB. To upload data that is greater than 20 MB, you must download the MaxCompute client and then use the tunnel command.
2. How to set the algorithm parameters?
To set the algorithm parameters, drag an algorithm component to the canvas and click the component. The corresponding parameters are displayed on the right-side pane.
3. How to view the experiment results?
If Machine Learning Platform For AI has successfully run a component, it marks the component with a green check. You can right-click a component with a green check to view data or evaluation results.
4. How to view and download the model generated from an experiment?
To generate a model, you must first select Setting > General > Auto-generate PMML from the left-side navigation pane. After successfully running an experiment, you can select Model from the left-side navigation pane to check the corresponding model. To view the model parameters, right-click the model. To download a model, right-click the model and select Download PMML.
5. What is PMML?
PMML is a standard model description file. A PMML file downloaded from Machine Learning Platform For AI can be applied to open-source engines, such as Spark.