liveBook · Manning

Customer identification

As explained in Fight Fraud with Machine Learning

Customer identification is the process of establishing and verifying a customer's identity using identifying information and documents.

Customer identification establishes whether a customer is who they claim to be. In online KYC, it compares user-entered identity information with evidence from an ID and selfie to prevent fraudulent onboarding. section 7.0 section 7.1.1

An automated identification workflow combines several inputs and checks:

  • Collect visual data such as a selfie and ID photograph, alongside textual data such as name, date of birth, and address. section 7.1.1
  • Extract information from the customer’s ID and compare it with information entered during onboarding. section 7.1 section 7.1.1
  • Run document quality, document fraud, and information-validation checks. section 7.4
  • Compare faces for duplicate-account detection or selfie-to-ID verification using face detection, embeddings, and cosine similarity. section 7.4

Selfie-to-ID verification

A customer submits a selfie and a Slovakian ID card. A face-detection model locates the face, a recognition model generates facial embeddings, and cosine similarity determines whether the images match. section 7.1.1 section 7.4

Figure 7.4 Different kinds of KYC data presented by a user including visual data such as selfie and ID photo, and textual data such as first name, last name, date of birth, and so on.

Use the data overview to inspect how selfies, ID photos, names, dates of birth, and other identity fields support automated verification. section 7.1.1

How Fight Fraud with Machine Learning covers it

How deep it goes
Explains identity inputs, ID information extraction, comparison, and face matching, while stopping short of detailed model training procedures.
Examples
Uses selfies, ID photographs, Slovakian ID cards, identity fields, duplicate checks, and selfie-to-ID comparisons.
What you can do after
Apply it via facenet_pytorch models to compare faces and verify whether a selfie matches an ID photograph.
How it connects
Connects KYC with customer due diligence, ongoing monitoring, document verification, face detection, face recognition, and deep learning.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Fight Fraud with Machine Learning. The section links point to the parts of the book it draws on.