The Next Applied
social share alt icon

IMPROVING TRUSTWORTHINESS OF AI MODELS

Know More

DEVELOPING UNDERSTANDABLE MODELS

 

AI/ML based solutions are becoming ubiquitous, solving many critical problems impacting people’s lives. Banks and financial institutions rely on AI based solutions to improve user experience and take lending decisions. Insurance companies use AI to identify policy claim frauds. Recruiters leverage them to identify the best candidates. With the pervasiveness and the black box nature of these complex models, there is growing need for models and solutions to be responsible and accountable.

Responsible AI develops artificial intelligence in a way that is inclusive and understandable by users. It provides explanations to AI models to improve trustworthiness, accountability, identification & mitigation of biases, and user and regulatory confidence. It allows the stakeholders to understand the system and assess how the model predictions change with different inputs. Regulators can check biased and discriminatory practices, and decision

OUR SOLUTION

 

Mphasis Responsible AI is an end-to-end framework that enables companies to develop and deploy robust, interpretable, explainable, bias-free, auditable, and privacy preserving AI through a unique user experience and design thinking engagement. This makes the system trustworthy, thereby improving customer experience, reducing liability risk, and ensuring regulatory compliance.

The Responsible AI components are generic and modular, enhancing scalability and repeatability across several use cases. The framework’s global and local explanations help in understanding internal logic and model limitations, and bias identification and mitigation modules assure model fairness. While a PII redaction algorithm can be leveraged to preserve privacy and logging of experiments, the model versions allow for explanation – accuracy trade-off analysis as well as auditability. Mphasis Responsible AI framework is fully integrated with PACE-ML, Mphasis’ proprietary MLOps framework for easy deployment.

 

KEY TENETS

 

High performing: Machine learning models achieve acceptable level of performance in terms of identified metrics (such as accuracy, sensitivity, specificity, F1 score, log-loss etc.)

Interpretable: A cause and effect can be observed at an extent that after some observations a human can predict the change in output, given a change in input of a working model

Explainable: The internal mechanics of a model can be explained in human terms

Auditable: Model actions and the attributes driving them are recorded with integrity and are readily available for scrutiny

Bias Free: The models are impartial; they work without giving undue advantage/disadvantage to any class

Privacy Preserving: The model is free from any PII data to respect privacy of all stakeholders

KEY FUNCTIONALITIES

 

Solution Design

  • User Experience & Design Thinking Workshops: Mphasis’ proprietary client engagement model leverages unique user journey mapping and design thinking workshops involving all the stakeholders, to create the most effective AI solution to address the root issues.
  • Gap Analysis: The gap-analysis / assessment is used to assess the maturity of the processes with respect to applicability of specific AI/ML solutions and the customizations.

Solution Design

Model Performance

  • Model Debugging: Performance is measured across a range of inputs to identify if there are any anomalies. Comparison with competing models or new scenarios can also be performed.
  • Monitoring: ML model performance metrics are continuously monitored to check for any degradation in performance. Change in model behavior from accepted standards need to be alerted for model refresh.
  • Data and Model Drift: Changes in data distributions and process/policy are captured though the drift analysis and incorporated back into the models through re-trainings, to maintain their performance.

Model Performance

Privacy Preserving

  • PII Protector: Several techniques are used to convert PII data into information that is anonymous and cannot be decoded by humans, keeping the mapping secure for future purposes.

Privacy Preserving

Bias Check

  • Bias Detection: Several techniques are used to find if model is positively or negatively biased towards a class. It can also be used to check fairness in data before initiating model building.
  • Bias Mitigation: Several human interventions and automated techniques are used to mitigate the biases pre-known or identified using bias. Bias mitigation can be carried out before, during and after initiating model development.

Bias Check

Interpretability & Explainability

  • Highlighting Key Features: The global explanation techniques explain how the model is doing predictions.
  • Decision Explanations: The local explanations techniques explain why a specific decision was taken for the given input.

Interpretability & Explainability

Auditability

  • Tracking: The complete lifecycle and components (data, code & models) of a ML project are versioned and tracked.
  • Logging: All the artifacts like metrics, explanations, predictions etc. are logged for post-hoc scrutiny and analysis throughout the ML project lifecycle (development, deployment and operationalization)
  • Governance: Methods which do not allow the tempering of audit trails, are deployed to keep track of how data is used or model is built for later scrutiny. This helps in comprehensive audibility and possibility of debugging any errors.

Auditability

BENEFITS

 

Improves the model outcomes: The framework allows users to check if AI predictions discriminate against a group. This helps developers mitigate the biased results and improve outcomes for the group in focus.

Ensures regulatory compliance: Responsible AI ensures increased AI transparency and helps understand reasons for a negative outcome.

Improves confidence in critical business decisions: Mphasis Responsible AI is useful when the model is taking critical decisions that impact the well-being of individuals.

Simplifies model governance: The auditability module makes the various steps in the model development track available for scrutiny through continuous model evaluation.

Increases User Trust: Mphasis Responsible AI produces human centric intuitive explanations, building user satisfaction and trust.

Eases deployment: The solution leverages pre-built components that offer faster deployment. The solution can be customized to suit varied business contexts and requirements.

Cloud-agnostic and hybrid friendly: The solution can be easily deployed on-prem as well on any private or public cloud.

Flexible integration: Responsible AI is deployable on its own (stand-alone) to add explanations to AI models already in production or can be integrated with other solutions.

SUCCESS STORIES

 

MPHASIS RESPONSIBLE AI SOLUTIONS
ON AWS MARKETPLACE

 

Through AWS Marketplace, Mphasis offers ML solutions that deliver immediate results and ROI in critical enterprise business processes and operations. These solutions can be deployed with the speed and security provided by AWS. Our presence in the AWS marketplace has facilitated the expanding global footprint of Mphasis AI offerings, helping developers get hands-on experience on AI algorithms and layer AI-based solutions in enterprise applications. Combined with AWS services, these ML solutions help in simplifying data experimentation, extract deeper insights from data estates and improve productivity for a variety of use cases.

Our current Responsible AI listings

   THOUGHT LEADERSHIP

 

Responsible AI in Large Scale Machine Learning System

 

Explainable AI for Mitigating Biases in Judgment Systems