AI Action Proctoring for Remote Skills Exam Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Challenges exist in accurately and objectively evaluating remotely located candidates during skills-based examinations that require performance of human actions, as existing proctoring systems struggle to maintain examination integrity and prevent cheating.
Innovation Solution
A method is provided for training and deploying an AI neural network to proctor skills-based examinations, using a hybrid CNN-LSTM network to analyze timestamped images from multiple vantage points, generating a candidate performance profile based on spatial and temporal features of human actions, and validating the network with training and accuracy functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional proctoring systems are used for remotely located candidates, then examination administration is simplified, but measurement precision of human actions deteriorates
Solution Approach 1:
The patent replaces traditional mechanical/proctor-based evaluation systems with an AI neural network system that uses computer vision and deep learning algorithms to automatically detect, track, and evaluate human actions. The neural network processes video feeds and sensor data to objectively measure candidate actions, substituting human proctor judgment with automated computational analysis.
Solution Approach 2:
The patent introduces an AI neural network as an intermediary between the candidate's physical actions and the examination evaluation system. This intermediary layer processes raw video and sensor data, extracts meaningful action patterns, and provides structured evaluation metrics, bridging the gap between physical human actions and digital assessment requirements.
2Reliability
If attempts to circumvent mandated procedures are prevented, then examination integrity is improved, but device complexity increases
Solution Approach 1:
The patent implements preliminary action by training the AI neural network in advance with extensive datasets of legitimate and fraudulent behaviors. The system pre-learns patterns of examination integrity violations and is ready to detect them during the actual examination. This preliminary training phase enables the system to proactively identify and prevent cheating attempts without requiring complex real-time intervention mechanisms.
Solution Approach 2:
The patent incorporates feedback mechanisms where the AI neural network continuously monitors candidate behavior, compares it against learned patterns, and provides real-time feedback signals. When potential violations are detected, the system generates alerts and can automatically adjust examination parameters, creating a closed-loop system that actively maintains integrity through continuous monitoring and response.
3Measurement precision
If AI neural network is deployed for action detection, then measurement precision of human actions is improved, but difficulty of detecting and measuring increases
Solution Approach 1:
The patent applies segmentation by breaking down complex human actions into smaller, more manageable action primitives or atomic movements. The AI neural network is trained to recognize these basic units separately, then combines them to understand complex sequences. This segmentation approach transforms the difficult problem of detecting entire complex actions into the more tractable problem of detecting and sequencing simpler movement elements.
Solution Approach 2:
The patent moves the detection problem from spatial analysis alone to include temporal dimensions by using video sequences and time-series data. The AI neural network analyzes actions across multiple time steps, capturing motion dynamics, velocity, acceleration, and temporal patterns. This dimensional expansion from static image analysis to dynamic video analysis provides additional information for more accurate detection of complex actions.
Data Source
AI summary
A method and system of training and deploying an artificial intelligence (AI) neural network for skills based examinations. A method of training the AI neural network includes providing, via input layers of the AI neural network, a training dataset of human action images for skills based examinations that require human actions performed in accordance with a predetermined sequence, generating, at an output layer of the AI neural network, a correlation between the training dataset and a validation dataset of human action images for the skills-based examination, the input layers and the output layer being interconnected via a set of fully connected layers of the AI neural network, and validating the AI neural network based on a training loss function expressed in accordance with the correlation between the training dataset and the validation dataset.