Computer Vision is the process of using statistical algorithms to automatically identify or verify an object in a digital image or a video frame. Math & Pencil has extensive experience with facial recognition applications - specifically, designing algorithms that are able to detect microexpressions on a human face. Utilizing the latest feature extraction techniques, such as wavelets & time series analysis, combined with sophisticated machine learning algorithms (both Bayesian and Deterministic) Math & Pencil has successfully constructed algorithms which determined the emotion exhibited on a human face.
Math & Pencil has been tasked with segmenting customer comments (text) based on semantic information. We have experience building semantic analysis engines which may be used to moderate, segment or flag different types of text based comments (from blog posts to comments on a website). Custom automation tools can assist the process of online comment moderation.
Recommendation engines are ubiquitous in todays online experience. Popular sites such as Netflix, Amazon, and Spotify pump smart recommendations at you on an hourly basis, and for a good reason - good recommendations increase revenue. We have been tasked with building a large scale recommendation engine from trading/transaction data for a large United States financial institution to simple recommendations on a dual core laptops which were scaled to large computing clusters using Apache Hadoop / Mahout and Amazon EC2.
Training algorithms that understand written user input and formulate intelligent responses has a wide of business applications. Math and Pencil has experience training NLP algorithms that power chat bots in a range of business applications with interfaces such as email, SMS, and the web.