Machine Learning and Artificial Intelligence (ML & AI) Software and Services

Applying state of the art ML/AI techniques to train, guide and monitor humans.

Intelligence is the next frontier for value creation. We help customers leverage the latest advances in machine learning and artificial intelligence to unlock new value. We apply state of the art machine learning and artificial intelligence techniques to solve problems in training, guidance and monitoring. Most frequently used techniques include deep neural network models for computer vision and natural language processing. Our team also applies a number of other traditional ML techniques as required by the problem statement. 

Over time, we have incorporated a number of ML/AI capabilities into HyperSkill, so that non-experts can take advantage of these capabilities without having to invest time to learn the state of the art methods. Key HyperSkill AI capabilities include conversational AI and computer vision based quality monitoring. To learn more about these capabilities, please visit the HyperSkill page

Our capable and highly motivated engineers are ready to assist both in-house as well as on-site, depending on customer requirements. Our team’s mission is to build immersive, interactive and intelligent software systems that can assess people’s knowledge, skills, affective and physiological states and personalize the experience to foster greater learning and performance. The following are some of the areas where we offer expert consulting services. Please contact us to learn more.

Computer vision applications

We are applying computer vision techniques to build powerful augmented reality (AR) applications for manufacturing and healthcare use cases which require fast and accurate detection, recognition and tracking of objects and actions in a variety of environments, with constraints on computational resources and power consumption. We use both fixed and mobile cameras in conjunction with head mounted displays (HMD’s) and mobile devices to solve customers’ problems.  For more details, please watch our recent presentation at Nvidia GTC held in March 2021

Natural Language Processing (NLP)

We apply NLP techniques for building natural user interfaces in AR, VR and mobile applications. For example, conversational agents in Virtual Reality simulations can provide deeper engagement, richer assessment, deeper learning, greater transfer and longer retention. Voice assistants in Augmented Reality applications can help with getting tasks done in areas such as training, selling, product assembly and installation, design reviews and even shopping. We have successfully deployed models for intent classification, Question answering, sentiment analysis and other applications. Depending on the customers’ needs, we select the best fit NLP technique from a range of options that include traditional methods to recent deep learning based models including foundation models such as GPT-3. If you are interested in creating your own conversational AI powered immersive simulations, please visit the HyperSkill page. We are also available to discuss custom solutions for your specific needs.

Bayesian network models

Machine learning for assessment of knowledge and skills

Bayesian Network models have been used in assessment to generate fine grained assessment for both summative and formative use. SimInsights has worked with academic researchers to apply Bayesian Networks and machine learning to assess knowledge and skills from observations of human performance in serious games and simulations. This multi-year collaboration has resulted in multiple research reports and software modules of increasing generality that can be applied to a range of inference problems. In addition to Bayesian methods, our team has also applied a variety of other techniques to assessment including Gaussian Mixture Models (GMM), clustering,  hidden markov, markov chains and traditional item response theory (IRT) based methods.  

Machine learning based behavior analytics

VR literature suggests a causal chain from immersion via presence to performance (Cummings, et al, 2012) which offers unprecedented power for assessing human responses to diverse stimuli. One example is human factors evaluation of new designs such as a hospital facility. 

However, the rich datasets pose a formidable challenge to human factors analysts. Manual coding of behaviors can be challenging and time consuming. Our team addresses this challenge by automating the analysis. We apply machine learning techniques to automatically recognize and classify complex behaviors and thereby speed up data analysis by factors of 10-100 times. By the time the participant takes off the VR headset, the data analysis is complete and the report is ready. Please contact us to learn more. 

Our unique approach to intelligence development