ONNX
morphish / README.md
eric707's picture
Create README.md
28f111e verified
|
raw
history blame
7.49 kB

Phishing Detection

Model Overview

Description:

  • Phishing detection is a binary classifier differentiating between phishing/spam and benign emails and SMS messages. This model is for demonstration purposes and not for production usage.

References(s):

Model Architecture:

Architecture Type:

  • Transformers

Network Architecture:

  • BERT

Input: (Enter "None" As Needed)

Input Format:

  • Evaluation script downloads the smsspamcollection.zip and extract tabular information into a dataframe

Input Parameters:

  • SMS/emails

Other Properties Related to Output:

  • N/A

Output: (Enter "None" As Needed)

Output Format:

  • Binary Results, Fraudulent or Benign

Output Parameters:

  • N/A

Other Properties Related to Output:

  • N/A

Software Integration:

Runtime(s):

  • Morpheus

Supported Hardware Platform(s):

  • Ampere/Turing

Supported Operating System(s):

  • Linux

Model Version(s):

  • v1

Training & Evaluation:

Training Dataset:

Link:

Properties (Quantity, Dataset Descriptions, Sensor(s)):

  • Dataset consists of SMSs

Evaluation Dataset:

Link:

Properties (Quantity, Dataset Descriptions, Sensor(s)):

  • Dataset consists of SMSs

Inference:

Engine:

  • Triton

Test Hardware:

  • DGX (V100)

Ethical Considerations:

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. For more detailed information on ethical considerations for this model, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards below. Please report security vulnerabilities or NVIDIA AI Concerns here.

Subcards

Model Card ++ Bias Subcard

What is the language balance of the model validation data?

  • English

What is the geographic origin language balance of the model validation data?

  • UK

Individuals from the following adversely impacted (protected classes) groups participate in model design and testing.

  • Not Applicable

Describe measures taken to mitigate against unwanted bias.

  • Not Applicable

Model Card ++ Explainability Subcard

Name example applications and use cases for this model.

  • The model is primarily designed for testing purposes and serves as a small pre-trained model specifically used to evaluate and validate the phishing detection pipeline. Its application is focused on assessing the effectiveness of the pipeline rather than being intended for broader use cases or specific applications beyond testing.

Fill in the blank for the model technique.

  • This model is designed for developers seeking to test the phishing detection pipeline with a small pre-trained model.

Name who is intended to benefit from this model.

  • The intended beneficiaries of this model are developers who aim to test the performance and functionality of the phishing pipeline using synthetic datasets. It may not be suitable or provide significant value for real-world phishing messages.

Describe the model output.

  • This model output can be used as a binary result, Phishing/Spam or Benign

List the steps explaining how this model works.

  • A BERT model gets fine-tuned with the dataset and in the inference it predicts one of the binary classes. Phishing/Spam or Benign.

Name the adversely impacted groups (protected classes) this has been tested to deliver comparable outcomes regardless of:

  • Not Applicable

List the technical limitations of the model.

  • For different email/SMS types and content, different models need to be trained.

Has this been verified to have met prescribed NVIDIA standards?

  • Yes

What performance metrics were used to affirm the model's performance?

  • F1

What are the potential known risks to users and stakeholders?

  • N/A

Link the relevant end user license agreement

Model Card ++ Saftey & Security Subcard

Link the location of the training dataset's repository.

Is the model used in an application with physical safety impact?

  • No

Describe life-critical impact (if present).

  • None

Was model and dataset assessed for vulnerability for potential form of attack?

  • No

Name applications for the model.

  • The primary application for this model is testing the Morpheus phishing detection pipeline

Name use case restrictions for the model.

  • This pretrained model's use case is restricted to testing the Morpheus pipeline and may not be suitable for other applications.

Name target quality Key Performance Indicators (KPIs) for which this has been tested.

  • N/A

Is the model and dataset compliant with National Classification Management Society (NCMS)?

  • No

Are there explicit model and dataset restrictions?

  • No

Are there access restrictions to systems, model, and data?

  • No

Is there a digital signature?

  • No

Model Card ++ Privacy Subcard

Generatable or reverse engineerable personally-identifiable information (PII)?

  • None

Protected classes used to create this model? (The following were used in model the model's training:)

  • N/A

Was consent obtained for any PII used?

  • N/A

How often is dataset reviewed?

  • Unknown

Is a mechanism in place to honor data subject right of access or deletion of personal data?

  • N/A

If PII collected for the development of this AI model, was it minimized to only what was required?

  • N/A

Is data in dataset traceable?

  • N/A

Are we able to identify and trace source of dataset?

  • N/A

Does data labeling (annotation, metadata) comply with privacy laws?

  • N/A

Is data compliant with data subject requests for data correction or removal, if such a request was made?

  • N/A