NOVA means "the NEW" It provides the capability to train and predict frameworks on deep learning and AI very easily with only RestAPI(Json).
- Only RestAPI
- Easy Connect to AI/DNN ! @Created Simple Framework for Easy
Note
Python Example -> https://github.com/lastime1650/VATEX_NOVA_AI/blob/main/api_request_client_sample.py
This guide provides instructions and examples for using the VATEX NOVA AI API for Deep Learning and Machine Learning tasks.
Before making requests, configure your client with the server's IP and port.
- {{ Sample Connection }}
- IP Address:
192.168.1.205 - Port:
10302 - Base URL:
http://192.168.1.205:10302
| Task | Method | Path |
|---|---|---|
| Deep Learning Train | POST |
/api/solution/util/nova/DL/train |
| Deep Learning Predict | POST |
/api/solution/util/nova/DL/predict |
| Machine Learning Train | POST |
/api/solution/util/nova/ML/train |
| Machine Learning Predict | POST |
/api/solution/util/nova/ML/predict |
| Task | Method | Path |
|---|---|---|
| WithId Samples Appending | POST |
/api/solution/util/nova/with_id/sample/push |
| WithId Sample y Edit | POST |
/api/solution/util/nova/with_id/sample/y/edit |
| WithId Sample X Edit | POST |
/api/solution/util/nova/with_id/sample/x/edit |
| WithId Sample Remover | POST |
/api/solution/util/nova/with_id/sample/remove |
| WithId - Machine Learning Train | POST |
/api/solution/util/nova/with_id/ML/train |
| WithId - Machine Learning Predict | POST |
/api/solution/util/nova/with_id/ML/predict |
| WithId - Deep Learning Train | POST |
/api/solution/util/nova/with_id/DL/train |
| WithId - Deep Learning Predict | POST |
/api/solution/util/nova/with_id/DL/predict |
"WithId" API added.
This finally eliminates the need for the Client to store samples directly in the local environment! Now, using the "WithId" API, you can add samples to the NOVA AI in advance, and build AI modeling at any time.
Note
WithId API - Python Example -> https://github.com/lastime1650/VATEX_NOVA_AI/blob/main/WithId_sample.py
This example demonstrates how to train a neural network for a multi-class classification task.
Endpoint: POST /api/solution/util/nova/DL/train
Request Body:
{
"id": "classification-model-01",
"data": {
"X": {
"source": [
[10, 20, 30], [11, 22, 33], [90, 80, 70], [95, 88, 77],
[50, 50, 50], [45, 55, 60], [12, 25, 35], [92, 81, 74]
]
},
"y": {
"source": [ "A", "B", "C", "B", "C", "A", "C", "B" ],
"y_type": "one_hot_encode"
}
},
"train": {
"layer": {
"input": { "shape": [3] },
"hiddens": [
{ "type": "dense", "units": 16, "activation_func": "relu" }
],
"output": { "units": 3, "activation_func": "softmax" }
},
"compile": {
"optimizer_func": "adam",
"loss_func": "categorical_crossentropy",
"metrics": ["accuracy"]
},
"fit": {
"epochs": 10,
"batch_size": 1,
"validation_split": 0.25
}
}
}Response: The response is a list of JSON objects, one for each epoch, containing the training and validation metrics.
[
{
"accuracy": 0.3333333432674408,
"loss": 1.0891636610031128,
"val_accuracy": 0.5,
"val_loss": 0.9032398462295532
},
{
"accuracy": 0.3333333432674408,
"loss": 1.082886815071106,
"val_accuracy": 0.5,
"val_loss": 0.9078031778335571
},
...
]Use the id of the trained model to make predictions on new data.
Endpoint: POST /api/solution/util/nova/DL/predict
Request Body:
{
"id": "classification-model-01",
"data": {
"X": {
"source": [
[11, 22, 33]
]
}
}
}Response:
The output provides the class with the highest probability (argmax), the lowest (argmin), and the probabilities for all classes (all).
{
"status": true,
"output": {
"argmax": { "B": 0.33860063552856445 },
"argmin": { "A": 0.32945770025253296 },
"all": {
"A": 0.32945770025253296,
"B": 0.33860063552856445,
"C": 0.3319416642189026
}
}
}This example shows how to train a neural network for a regression task to predict a continuous value.
Endpoint: POST /api/solution/util/nova/DL/train
Request Body:
{
"id": "regression-model-01",
"data": {
"X": {
"source": [
[10, 20, 30], [11, 22, 33], [90, 80, 70], [95, 88, 77],
[50, 50, 50], [45, 55, 60], [12, 25, 35], [92, 81, 74]
]
},
"y": {
"source": [ 1.0, 2.0, 3.0, 2.0, 3.0, 1.0, 3.0, 2.0 ],
"y_type": "min_max"
}
},
"train": {
"layer": {
"input": { "shape": [3] },
"hiddens": [
{ "type": "dense", "units": 16, "activation_func": "relu" }
],
"output": { "units": 1, "activation_func": "sigmoid" }
},
"compile": {
"optimizer_func": "adam",
"loss_func": "mse",
"metrics": ["mae"]
},
"fit": {
"epochs": 10,
"batch_size": 1,
"validation_split": 0.25
}
}
}Response: The API returns a list of metrics for each training epoch.
[
{
"loss": 0.2034807652235031,
"mae": 0.38669994473457336,
"val_loss": 0.13519415259361267,
"val_mae": 0.3201397657394409
},
{
"loss": 0.20247896015644073,
"mae": 0.38515543937683105,
"val_loss": 0.13282975554466248,
"val_mae": 0.3143080770969391
},
...
]Endpoint: POST /api/solution/util/nova/DL/predict
Request Body:
{
"id": "regression-model-01",
"data": {
"X": {
"source": [
[11, 22, 33]
]
}
}
}Response: The output is the single predicted numerical value.
{
"status": true,
"output": {
"output": 2.010890245437622
}
}Endpoint: POST /api/solution/util/nova/ML/train
Request Body:
{
"id": "MyML-01",
"data": {
"X": { "source": [
[10, 15, 20], [12, 18, 25], [9, 14, 19], [11, 17, 23],
[50, 55, 60], [53, 58, 63], [47, 52, 57], [55, 60, 65],
[90, 85, 80], [88, 82, 78], [92, 87, 83], [95, 89, 84]
]
},
"y": {
"source": [ "A", "A", "A", "A", "B", "B", "B", "B", "C", "C", "C", "C" ],
"y_type": "label"
}
},
"train": {
"model": {
"model_name": "RandomForestClassifier",
"model_params": { "n_estimators": 100, "random_state": 4 }
},
"trainset": { "test_size": 0.25, "shuffle": true, "stratify": "y" }
}
}Response:
The output field represents the model's accuracy score on the test set.
{
"status": true,
"output": 1.0
}Endpoint: POST /api/solution/util/nova/ML/predict
Request Body:
{
"id": "MyML-01",
"data": {
"X": { "source": [ [10, 15, 20] ] }
}
}Response: The response includes the predicted probabilities for each class.
{
"status": true,
"output": {
"argmax": { "A": 0.98 },
"argmin": { "C": 0.0 },
"all": { "A": 0.98, "B": 0.02, "C": 0.0 }
}
}Endpoint: POST /api/solution/util/nova/ML/train
Request Body:
{
"id": "Regression-01",
"data": {
"X": { "source": [[1], [2], [3], [4]] },
"y": { "source": [2.0, 4.1, 6.0, 8.2], "y_type": "raw" }
},
"train": {
"model": { "model_name": "LinearRegression" },
"trainset": { "test_size": 0.25, "shuffle": true }
}
}Response:
The output field represents the Mean Squared Error (MSE) on the test set. A lower value indicates better performance.
{
"status": true,
"output": 0.0
}Endpoint: POST /api/solution/util/nova/ML/predict
Request Body:
{
"id": "Regression-01",
"data": { "X": { "source": [[5]] } }
}Response: The response provides the predicted value(s), their mean, and standard deviation.
{
"status": true,
"output": {
"predicted_values": [10.199999999999998],
"mean": 10.199999999999998,
"std": 0.0
}
}Clustering is an unsupervised task, so only input data X is required.
Endpoint: POST /api/solution/util/nova/ML/train
Request Body:
{
"id": "Clustering-01",
"data": { "X": { "source": [[1, 2], [1, 4], [5, 6], [6, 5]] } },
"train": {
"model": {
"model_name": "KMeans",
"model_params": { "n_clusters": 2, "random_state": 42 }
}
}
}Response:
The output field returns a performance score for the clustering model (e.g., silhouette score).
{
"status": true,
"output": 0.591051482474553
}Predict the cluster assignment for new data points.
Endpoint: POST /api/solution/util/nova/ML/predict
Request Body:
{
"id": "Clustering-01",
"data": { "X": { "source": [[2, 3], [5, 5]] } }
}Response: The response details the cluster assignments for the input data.
clusters: An array mapping each input data point to a cluster index.unique_labels: A list of all unique cluster labels.cluster_counts: A dictionary showing how many input points were assigned to each cluster.
{
"status": true,
"output": {
"clusters": [0, 1],
"unique_labels": [0, 1],
"cluster_counts": { "0": 1, "1": 1 }
}
}