Google Machine Learning Drawing Tool

Machine Learning by Examples

Automobile Learning by Examples victimisation Scikit-Learn, Keras, TensorFlow, PyTorch, and OpenCV.

1. Google Colab Notebooks (Relinquish Nvidia Tesla K80 GPU)
2. Sketcher using Keras/TensorFlow and QuickDraw-Dataset
3. Disease-Prediction victimization Machine Learning (Scikit-Learn)
4. Recruitment Co-ordinated exploitation Motorcar Learning (Keras & Tesorflow)

1.1. Configuring Evolution Environment using Google Colab Notebooks

• Step 1. Creating Folder happening Google Drive or pick out the default Colab Notebooks folder
• Footstep 2. Opening or Creating a `Colab Notebook
  • Openning
  • Google Drive: upload onto My Drive/machine-learning
  • Google Driveway: My Drive/machine-learning/disease-diagnostic-from-symptoms/disease_symptoms_data_analysis_DecisionTree.ipynb > Open with > Colaboratory
  • Creating new Colab Notebook computer via Right click > More than > Colaboratory
• Step 3. Stage setting Free GPU:
  • Google Colab: Edit > Notebook settings:
    • Runtime type: Python 3
    • Hardware accelerator: GPU
• Running Oregon Importing Files with Google Colab

                                        from Google.colab significance driveway     drive.mount("/content/gdrive", force_remount=True)                                  

  Note: Click the radio link, copy substantiation write in code and paste it to text boxful; then we can use /content/gdrive/My Drive/
• Install Python Module/Package

                                        # Install Excel/GoogleSheet Python mental faculty     !pip3 install --climb -q gspread     !pip3 install --upgrade -q xlrd          # Install Keras     !pip3 install -q keras     !pip3 install woolly mullein torchvision                                  

• Google Colab Notebooks

                                        # RAM & CPU     !cat /proc/meminfo     !cat /proc/cpuinfo     # Restart Google Colab     !kill -9 -1                                  

• Preeminence: 12-hour GPU limit is for a continuous assignment of VM.

1.2. Usefull Utilities

1.2.1. Facets

1.2.2. Tensorboard

• Upload file using gsutil dominate to GCS(Google Cloud Storage)

                    # First, we need to put up our project. Replace the assignment below #with your project Idaho. project_id = 'chatbotdemo-ai' !gcloud config solidifying plan {project_id} import uuid # Make a unique bucket to which we'll upload the file. # (GCS buckets are partly of a single globular namespace.) bucket_name = 'sample-bucket-' + str(uuid.uuid1())  # Full reference: https://mist.google.com/warehousing/docs/gsutil/commands/mb !gsutil mb gs://{bucket_name} # Copy the data file to our sunrise bucket. # Full reference: https://cloud.google.com/storage/docs/gsutil/commands/cp !gsutil cp trained_model.pkl gs://{bucket_name}/                                  

• Upload file from google drive to GCS(Google Cloud Storehouse)

                                        This subdivision demonstrates how to upload files using the native Python API rather than gsutil.     This snippet is based happening a larger lesson with additional uses of the API     # The first step is to create a bucket in your cloud project.     # Replace the assignment below with your cloud project ID.     # For details on cloud projects, see:     project_id = 'chatbotdemo-ai'     # Authenticate to GCS.     from google.colab import auth     auth.authenticate_user()      # Create the service client.     from googleapiclient.discovery import build     gcs_service = build('storage', 'v1')      # Generate a random bucket name to which we'll upload the file.     import uuid     bucket_name = 'sample-bucket-' + str(uuid.uuid1())      dead body = {     'name': bucket_name,     # For a wide-cut heel of locations, see:     # https://cloud.google.com/depot/docs/bucketful-locations     'location': 'us',     }     gcs_service.buckets().sneak in(project=project_id, consistency=body).execute()                                  

• Download file using gsutil command on GCS(Google Overcast Computer memory)

                                        # Download the Indian file.     !gsutil cp gs://{bucket_name}/trained_model.pkl /tmp/trained_model.pkl          # Print the result to construct sure enough the transfer worked.     !cat /tmp/trained_model.pkl                                  

• Download file from google drive to GCS(Google Cloud Store)

                                        We repeat the download instance above victimisation the native Python API.     # Authenticate to GCS.     from google.colab import auth     auth.authenticate_user()      # Create the servicing customer.     from googleapiclient.discovery import build     gcs_service = build('store', 'v1')      from apiclient.hypertext transfer protocol import MediaIoBaseDownload      with open('/content/gdrive/My Drive out/trained_model.pkl', 'Wb') as f:     request = gcs_service.objects().get_media(bucket=bucket_name,                                                 objective='trained_model.pkl')     media = MediaIoBaseDownload(f, request)      done = False     while not done:         # _ is a procurator for a progress object that we ignore.         # (Our file is small, so we skitter reporting progress.)         _, cooked = media.next_chunk()                                  

2. Sketcher using Keras/TensorFlow and QuickDraw-Dataset

A simple tool that recognizes drawings and outputs the names of the rife drawing. We will use Google Colab for preparation the model, and we bequeath deploy & run straight on the browser using TensorFlow.js.

2.1. Dataset

We will use a CNN to recognize drawings of unlike types. The CNN will equal trained happening the Quick-Draw Dataset.

2.2.

2. Disease-Prediction exploitation Machine Learning (Scikit-Watch)

3. Recruitment Twinned using Automobile Encyclopedism (Keras &adenosine monophosphate; Tesorflow)

References

• FastAI

Google Machine Learning Drawing Tool

Source: https://github.com/nnthanh101/machine-learning-by-examples

Share this post