Platform
Go to Platform
How it works
Engineer
Data scientist
Developer
Team lead
Features
Solutions
Go to Solutions
Our offer
Industry
Chatbots
Image similarity
NLP
Sentiment analysis
Speech to text
Integrations
Stories
Use cases
Learn
Go to Learn
Knowledge center
Documentation
Tutorials
Webinars
Resources
Blog
Go to Blog
Data science
Product development
Applied AI & AI in business
VLOG
About us
Go to About us
Join us
Press releases
Partner network
Pricing
TRY FOR FREE
Log in
Knowledge Center
Home
Documentation
Get started with the platform
Video walkthrough
Structure of this guide
Before we begin...
Platform overview and general workflow
Projects view
Datasets view
Modelling view
Evaluation view
Deployment view
Build, train, and deploy models - detailed workflow
Step 1 - Create a Project
Step 2 - Upload your data
Step 3 - Configure your dataset
Step 4 - Build a model
Step 5 - Train a model
Step 6 - Evaluate your model
Step 7 - Deploy your model
Use your deployed models in your applications
Using the Deployment API
Using Sidekick
Bubble integration - No-code apps
Learn by doing
Next step
Tutorials
For beginners
For intermediate users
For expert users
Deploy an operational AI model
Create a project
Add the MNIST dataset to the platform
Design a deep learning network with the wizard
Run experiment
Analyze experiment
Deploy your trained experiment
Test the MNIST classifier
Tutorial recap and next steps
Predict real estate prices
Create a project
The data
Add the Calihouse dataset to the platform
Design a model for image data
Create experiment
Run experiment
Run concurrent experiment with a second input for tabular data
Analyze experiments
Deploy trained experiment
Tutorial recap
Self sorting wardrobe
The problem
Create project
Add and manage the dataset
Select a CNN model
Run experiment to train model
More experiments
Analyze experiments
Tutorial recap
Next steps / Read more
Movie review feelings
Create a project
What is BERT?
Add the data
BERT - Design a text classification model
Create experiment
Check experiment settings & run the experiment
Analyzing the first experiment
Deploy your trained experiment
Test the text classifier in a browser
Test the text classifier in a terminal
Tutorial recap and next steps
Classify text in any language
The data – books in many languages
Create a project
Import the data
Multilingual BERT - the text processing AI
Create the experiment
Run the experiment
Evaluate your model
Enable the deployment
Test the deployment
Recap
Build your own music critic
Create project - Figure out the mood of a song
Upload dataset
Preprocess the dataset on platform
Select DenseNet in the Experiment wizard
Start training from the modeling view
Run concurrent experiment
Analyze experiments
Last iteration with the whole dataset
Deploy your model
Tutorial recap - You have solved a real-world problem
Related content
Use AI to detect fraud
The problem
Credit card fraud data
Add class weights and run the model
Run experiment
Run several experiments
Evaluate
Next step on the platform
Further reading
Look deep into DNA
Motivation - Replicate results of a research paper
Biological background
Formulating the problem in terms of deep learning
Work with the data on the platform
Train the model
Conclusion
Skin cancer detection
The problem - Predict lesion segmentation boundaries
The data
Goals of the experiment
Create project
Add the dataset
Create a model for image segmentation
Train the model
Analyze experiment
Test the model
Tutorial recap
Car damage assessment
Create project
Add dataset to the project
Build a model in the Experiment wizard
Run experiment
Analyze the experiment
Deploy the trained model
Tutorial recap
Further reading
Verify images with Zapier and Peltarion
Create a spreadsheet with insurance claims
Create Zapier flow: Peltarion <--> Google Sheet
Recap
Next project
Kaggle competition with zero code
Preprocessed data
Create a new project
Add dataset to the platform
Create a deep learning experiment
Run experiment to train the model
Analyze experiment
Download model
Getting started with Kaggle account
Submitting predictions
Further work — Transfer learning
Writing style tutor
Test the web app first
You will learn to
The data – 100 most downloaded ebooks
Build a model on the platform
Load a BERT snippet
Create experiment
Run experiment
Enable the deployment
Create a web app – use the model
Test the classifier in a browser
Share your results
Recap
Next steps and linked ideas
Find similar images of fruits
Create a project
Add the Fruit 360 dataset to the platform
Build model with the Experiment wizard
Change to only 1 epoch
How does image similarity search work?
Deploy model
Test deployment
Tutorial recap
Further reading
Test image similarity deployment with Postman
Denoising images
The problem — Denoising images
The data
Create a project for denoising images
Add the grayscale MNIST dataset to the platform
Design a deep learning autoencoder
Run experiment
Analyze experiment
Test if your autoencoder can remove noise
Tutorial recap
Alternative solutions to image denoising
Shape up your Slack chaos
You will learn to
Start here
Create a Slack app that can collect your conversation history
The Data – Scrape your conversation history
Build, train and deploy a model on the platform
Set the Slack bot live in a #channel
Run your Slack bot with Docker
Let the bot into your Slack
Deploy your besserwisser bot in Google Cloud
Book genre classification
Preread
Embeddings
The problem - sci-fi or centaurs?
Goal with this experiment
Dataset - CMU book summary dataset
Create a project
Add the data
BERT - Design a text binary classification model
Create experiment
Check experiment settings & run the experiment
Analyzing the first experiment
Deploy your trained experiment
Test the text classifier in a browser
Tutorial recap and next steps
Audio analysis for industrial maintenance
The problem
You will learn to
The data
Create a project
Build the model
Run the experiment
Evaluating the experiment
Deploy your model
Recap
Create a no-code AI app
Use our deployed AI model or create your own
Create a new app with Bubble
Add the Peltarion plugin to Bubble
Design the app
Create the workflow
Extract data from the AI model
Use predictions in app
Test the app
Ideas for next steps
What you’ve learned
Understand the mood of your team with Slack data
Set up your data collection from Slack
Make a positivity prediction in Peltarion
Extract timestamp
Collect data in Google Sheets
Create dashboard in Google sheets
Final words
Sales forecasting with spreadsheet integration
Create a project
Import the data
Build the model
Evaluate your model
Deploy your model
Integrate your model where you work
Recap
Integrate AI into Microsoft Excel for sales forecasting
The data
Train a model
Install the AI for Excel add-in
Get AI predictions in Excel
Recap
Integrate AI into Google Sheets for sales forecasting
The data
Train a model
Install the AI for Sheets add-on
Get AI predictions in Google Sheets
Recap
Detecting defects in mass produced parts
Create a project
Import the data
Preprocess the data
Build the model
Run the experiment
Evaluate the experiment
Deploy your model
Buy or not / Predict from tabular data
The problem - Unleash the power of the spreadsheet
Getting started - create a project
The data
Build your model in the Experiment wizard
Evaluation view
Improve your model
Further reading
How to improve a model that uses tabular data
Run several experiments and test new ideas
Increase patience to train for more epochs
Change the model architecture
Change the learning rate
Set a learning rate schedule
Increase batch size
Classify customer complaints
Create a project
Download the data from CFPB
Upload and process the data
Create an experiment and build a model
Deploy the model
Next step
Use Peltarion connector in Microsoft Power Apps
Create your Power App
Add the Peltarion connector
Call the deployed Peltarion model
Run the app and get a prediction
What you've learned
Projects view
Create a new project
Navigate to a project
Recent experiments overview
Recent experiments menu
Search and filter for projects
Share your filters
Datasets view
Dataset overview
Dataset view for a specific dataset
Import files and data sources to the Platform
File import
Data library: ready-made datasets
Data warehouse: import datasets from Azure Synapse and BigQuery
Requirements of imported files
Zip file specifications
Csv file specifications
Images specifications
Npy file specifications
Data warehouse import: Azure, BigQuery
Azure Synapse import
BigQuery import
Edit an imported dataset for use in experiments
Edit and inspect datasets
Dataset features
Feature set
Feature encoding
Types of encoding
Categorical encoding
Binary encoding
Image encoding
Text encoding
Normalization
Subset of a dataset
How to create subsets
Filter data
Split subset
Feature distribution
Understanding the feature distribution
Subset statistics
Working with multiple dataset versions
Naming of dataset versions
How to check subset settings of a saved dataset version
How to edit a dataset version
Create a new version of a dataset
Search and filter for datasets
Datasets used in tutorials
Calihouse dataset
California housing
Open street map
Download complete dataset
Licence
Fashion MNIST dataset
Download here
Licence
MNIST dataset
Download here
Licence
Tagger dataset
Licence
The Large Movie Review Dataset
Content and format of the raw dataset
Data library dataset is preprocessed
Explore the dataset with Python
Terms of use
Bank marketing
Dataset origin
Attribute information
Example workflows - Datasets view
Create new dataset / Example workflow
Step 1: Data acquisition
Step 2: Data preprocessing
Step 3: Create csv file
Step 4: Create the zip-file
What’s next
Impact of standardization - create different versions of a dataset / Example workflow
Step 1: Create a new project
Step 2: Import data
Step 3: Rename a feature with a meaningful label
Step 4: Manage data encoding
Step 5: Create five versions of the dataset
Conclusion
Modeling view
Modeling canvas
Experiments overview
Build an AI model
Create ready-to-run experiment with the Experiment wizard
Create blank experiment
Add a block or a snippet to an experiment
Snippets - your gateway to deep neural network architectures
Choosing the right snippet
Going further
DenseNet snippet
Tiramisu snippet
U-net snippet
VGG snippet
Inception snippet
ResNetv2 snippet
CNN snippet
CNN + FC snippet
Pretrained snippets
Transfer learning with pretrained snippets
Available pretrained snippets
How to use pretrained snippets
VGG - pretrained
MobileNetV2 - pretrained
EfficientNet - pretrained
Multilingual BERT snippet
English BERT snippet
XLM-R Embedding snippet
Universal sentence encoder snippet
BERT - pretrained
Blocks
Input
Target
Output
Dense
Activation
Scaling
Dropout
Batch normalization
2D Convolution block
2D Deconvolution block
2D Depthwise convolution
2D Max pooling block
2D Average pooling
2D Global average pooling
2D Global max pooling
2D Upsampling
2D Zero padding
1D Average pooling
1D Convolution block
1D Global average pooling
1D Global max pooling
1D Max pooling block
1D Upsampling
1D Zero padding
Flatten
Concatenate
Add
Multiply
Reshape
Embedding
BERT Tokenizer
Multilingual BERT encoder
English BERT encoder
Universal sentence encoder
XLM-R Tokenizer
XLM-R Encoder
Text embedding
BERT Encoder
Image augmentation
Image augmentation settings
How to augment images in a dataset
Class weights
Why use class weights
How to use class weights
How does class weighting work
Test it on the platform
Activations
Linear
ReLU
Swish
Sigmoid
Softmax
Tanh
Hard sigmoid
Loss functions
Choosing a loss function
Compatibility with activation functions
Binary crossentropy
Categorical crossentropy
Mean absolute error
Mean squared logarithmic error (MSLE)
Mean squared error
Poisson
Squared hinge
Run a model
Before running a model
The Run settings section
Batch size
Epoch
Data access seed
Optimizers
The Run button
Early stopping
Why use early stopping
How to use early stopping
Optimization principles (in deep learning)
Optimization with labeled training
Model loss
Batch size
Learning rate
B and B2 rate
p
Momentum
Nesterov momentum
Resources
Learning rate schedule
The optimizers
Optimizer test with Rosenbrock function
How optimizers behave at a saddle point
Adadelta
Adagrad
Adam
Adamax
AMSgrad
Nadam
RMSprop
Stochastic gradient descent
Experiment states
Created state
Queued state
Running state
Paused state
Completed state
Failed state
Dequeued state
Improve your model
Change run settings
Change block parameters
Working with multiple experiments
How to duplicate experiment
Link to parent experiment
Copy blocks with weights to another model
Multiple inputs
Transfer learning
Trainable or non-trainable
Copy with weights parameters
How to copy blocks with weights
Download and deploy model with weights
h5 limitations
Search and filter for experiments
Share your filters
Experiment tagging
How to edit an experiment’s tags
How to use tags in the experiment search
Modeling canvas controls
Controls to navigate the Modeling canvas
Tools
macOS keyboard shortcuts
Windows/Linux keyboard shortcuts
Information pop-up
Troubleshooting
Example workflows - Modeling view
Modeling view - with and without standardization on image data / Example workflow
Step 1: Create experiments with dataset version [.userinput]#NoStdImage/TargetStd#
Step 2: Configure the dataset settings
Step 3: Configure the blocks settings in the CNN snippet
Step 4: Config the settings for running the model
Step 5: Click Run
Step 6: Duplicate the experiment
Step 7: Create experiments with dataset version [.userinput]#StdImage/TargetStd#
Conclusion
Modeling view - with and without standardization on tabular data / Example workflow
Step 1: Create experiments with dataset version [.userinput]#NoStdTabular/TargetStd#
Step 2: Configure the dataset settings
Step 3: Configure the blocks settings in the CNN snippet
Step 4: Config the settings for running the model
Step 5: Click Run
Step 6: Duplicate the experiment
Step 7: Create experiments with dataset version [.userinput]#StdTabular/TargetStd#
Conclusion
Evaluation view
Evaluation view elements
Loss and metrics
Navigating the loss and metrics plot
How to read the loss curve
Predictions inspection
Select the subset and checkpoint to inspect
Predictions table
Confusion matrix
Interact with the confusion matrix
How to improve classification results
Multi-dimensional target
Scatter plot
Interact with the scatter plot
How to improve regression results
Multi-dimensional target
ROC Curve
How to read the ROC curve
Interact with the ROC curve
How to decide on a threshold
Classification loss metrics
Loss curve
Binary accuracy
Suggestions on how to improve
Categorical accuracy
Suggestions on how to improve
Binary error
Suggestions on how to improve
Categorical error
Suggestions on how to improve
Precision
Definition
Macro-precision
Precision
Macro-averaging
Micro-precision
Precision
Micro-averaging
Recall
Definition
Macro-recall
Recall
Macro-averaging
Micro-recall
Recall
Micro-averaging
AUC / Area under curve
AUC explained
F1-score
Definition
Macro F1-score
Definition
Macro-averaging
Micro F1-score
Definition
Micro-averaging
Regression loss metrics
MAE / Mean absolute error
Definition
Why use MAE
When to use MAE
Example of use
MSE / Mean squared error
Definition
Why use MSE
When to use MSE
Example of use
RMSE / Root mean squared error
When to use RMSE
MAPE / Mean absolute percentage error
Example of use
R2 / R-squared
When to use R-squared
Measure performance when working with imbalanced data
Evaluate on macro-precision, -recall, and -F1
Evaluate with the confusion matrix
Don’t evaluate on loss or accuracy
Test it on the platform
Search and filter for experiments
Share your filters
Experiment tagging
How to edit an experiment’s tags
How to use tags in the experiment search
Example workflows - Evaluation view
Evaluation view - with and without standardization / Example workflow
Step 1: Document experiment results
Step 2: Compare results
Conclusion:
Deployment view
Enable deployment for requests
Parameters
API information
Using your deployment
Deploy to API limitations
Unsupported deployment scenarios
Request size
Latency
Input data
Search and filter deployed experiments
Share your filters
Organization settings view
Access Organization settings view
Members view
Add and remove members of the organization
Platform roles
Possible account membership status in the organization
How to remove a team member
How to change the role of a team member
How to invite a team member
How to resend a team member invitation
How to withdraw a new team member invitation
My profile view
Reset password
Leave organization
Quota view
Organization resources
How to manage quotas
End of organization quota plan
Cheat sheets
Problem types
Specific use cases
BERT - Text classification / cheat sheet
Use this cheat sheet
Data preparation
Modeling view
Evaluation view
Deployment view
Multi-label image classification / cheat sheet
Problem formulation
Data preparation
Modeling
Single-label image classification / cheat sheet
Use this cheat sheet
Data preparation
Modeling
Deployment view
Image segmentation / mark a single object type within an image / cheat sheet
Problem formulation
Data preparation
Modeling
Evaluation
No-code AI / cheat sheet
No-code connections
Start with an idea and collect data
Start with the AI model on the Pelation Platform
Use your trained AI model
Example: Mood dashboard
USE - Text similarity / cheat sheet
What does it mean?
How to use it?
Image similarity / cheat sheet
Problem formulation
Data preparation
Modeling
Evaluation
Deployment
Test the deployment
Snippet selector for image projects
Glossary
AI concepts
Available AI concepts
Image similarity
What can you do with image similarity?
Similarity implemented on the platform
Deep dive explanation of image similarity
Further reading
Peltarion APIs
What can you do with the Peltarion APIs?
How can you use the Peltarion APIs?
Data API
Enabling the Data API
Prerequisites
How to upload files
Examples
Deployment API
Prerequisites
Structure of a prediction request
Examples
Working with responses
Understanding prediction results
Test the deployment from a Python script
Python script example with deployed mnist model
More than one input
String and numerical values as input
AI for Excel
How to install Peltarion AI for Excel
How to use Peltarion AI for Excel
AI for Sheets
How to install Peltarion AI for Sheets
How to use Peltarion AI for Sheets
Test with Postman
Available how-tos
Test image similarity deployment
Deploy an experiment
Configure Postman
Manage environment
Test your deployment with an image
Show results
Download API spec
Terms
Terms of use
Privacy notice
Dataset licenses
Pretrained licenses
Dataset licenses
17k Mobile strategy games
License
Boat types
License
Book summaries
References
License
Calihouse
California housing
Open street map
How to credit OpenStreetMap
Download complete dataset
Car damage
License
Download dataset
Cifar-10
License
Download dataset
Deep learning yeast UTRs
References
License
DeepWeeds
References
License
Defects in metal casting
References
License
Fashion-MNIST
License
Download dataset
Fine food reviews
References
License
Flower photos
References
License
Freesound Audio Tagging
References
License
Fruits 360
References
License
German Traffic Sign Recognition Benchmark (GTSRB)
References
License
Grocery store
References
License
Imagenette
License
IMDB
References
License
Industrial machinery operating conditions
References
License
MNIST
License
Download dataset
News headlines
References
License
Most downloaded public domain books
License for mirror
Download dataset
Pneumonia detection
References
License
Oxford 102-category flowers
References
License
Oxford IIT Pet
References
License
PlantVillage
References
License
Pokemon images
License
Sign language for alphabets
License
Sign language MNIST
License
Skin lesion segmentation
References
License
Download dataset
Spoken verbs
License
Stack Overflow Tags
License
Stanford Online Products
References
License
Tagger
Download dataset
Tencent ML-Images
References
License
Pretrained licenses
Snippets with weights licenses
VGG
References
License
DenseNet
References
License
ResNet
References
License
MobileNetV2
References
License
EfficientNet
References
License
BERT
Bidirectional Encoder Representations from Transformers
References
License
XLM-R
XLM-RoBERTa
References
License
Universal sentence encoder
License
References
Troubleshooting
Error messages
Modeling view
Experiment
Known issues
FAQ
General questions
Datasets view
Modeling view
Deployment view
Subscription questions
Technical requirements
GitHub repositories
Sidekick
Community-code
Documentation
/
Troubleshooting
Troubleshooting
Error messages
Known issues
TRY FOR FREE