HOLY GRAIL
We use cookies to provide the best site experience.
Ok, don't show again
Super intelligence for research and optimization problems
SINCE 2042
Super intelligence for research and optimization problems
SINCE 2042
Why
Deep Learning methods can accelerate R&D and process optimization massively but they are too complex to implement and use
What
Deploy a Probabilistic Neural Network with active learning in less than 1 minute
Or with few lines of code
Rest API, available in python, cURL, C, C#, Java Script, and other languages
Python - Requests
import requests
url = "https://bayes-api.holygrail.ai:80/api/projects/autonomous?api_key=yourapikey"
payload = {'request': '{
"name": "yourdataset",
"targets": [
{
"name": "target1",
"type": "MAXIMIZE",
"priority": 10
},
{
"name": "target2",
"type": "MINIMIZE",
"priority": 8
},
{
"name": "target3",
"type": "VALUE",
"value": "27"
"priority": 6
}
],
"features": [{
"name": "feature1",
"type": "BINARY"
}, {
"name": "feature2",
"type": "CATEGORICAL",
"categorical": ["category1", "category2"]
},{
"name": "feature3",
"type": "NUMERICAL",
"numerical": {
"min": 1,
"max": 10,
"res": 0.5
}
}],
"batch_size": 20
}'}
files = [
('file', open('/C:/Users/user/Documents/Datasets/yourdataset.csv','rb'))
]
headers= {}
response = requests.request("POST", url, headers=headers, data = payload, files = files)
print(response.text.encode('utf8'))Under the hood
General Framework
Applicable to a wide range of problems where the goal is to predict or optimize a set or targets
Easy to use
Browser based interface or REST API
Feature engineering
The Brain accepts categorical and numerical features and targets and does automatic preprocessing
Model selection
The optimal model architecture is selected and benchmarked in each iteration
Large search spaces
Suitable for large combinatorial spaces where other methods are intractable
Multi-objective
Optimizes different components in a system using multiple features/inputs and targets/outputs
Hyper parameter tunning
The Brain finds the optimal parameters for different types of Neural networks and other probabilistic methods
Active learning
The Brain defines which experiments to execute next or data points to acquire for optimal knowledge gain
Under the hood
General Framework
Applicable to a wide range of problems where the goal is to predict or optimize a set or targets
Easy to use
Browser based interface or REST API
Feature engineering
The Brain accepts categorical and numerical features and targets and does automatic preprocessing
Model selection
The optimal model architecture is selected and benchmarked in each iteration
Large search spaces
Suitable for large combinatorial spaces where other methods are intractable
Multi-objective
Optimizes different components in a system using multiple features/inputs and targets/outputs
Hyper parameter tuning
The Brain finds the optimal parameters for different types of Neural networks and other probabilistic methods
Active learning
The Brain defines which experiments to execute next or data points to acquire for optimal knowledge gain
Industries
Energy
Chemicals
Cement
Manufacturing
Biology
Health
Materials
Others
Find out how we can help solve your complex problems
Get the Holy Grail updates