MPT4U.com

In 2009 MPT4U was launched. The founders, Javier R. Movellan, Stanley Kim, Gwen Littlewort Ford, announced themselves as "a new start up company dedicated to the development of Machine Perception Technologies for You." Machine Perception Technologies developed expression recognition applications that numerically represent human facial expressions. Their headquarters were in San Diego. For three years their domain was live and then it expired.

When I discovered that the domain was available I bought it with the goal of rebuilding the site from its archived content. I happen tp work for a progressive software company as part of a Salesforce development team. We build custom applications, as well as responsive Salesforce employee-facing mobile enterprise apps on Force.com for all sorts of businesses and organization. My team regularly works with such front-end technologies such as JavaScript, HTML5, CSS3, JQuery, ExtJS, Ajax, as well as software languages: Java, Grails, Groovy, and PHP. With my interests in the sciences & technology I have had a long standing facination with capability of a computer system to interpret data in a similar manner to the way humans use their senses to relate to the world around them. Thus my interest in what this San Diego company was doing.

Consider this site as a historical documentation of the company known as Machine Perception Technologies, which is now closed.

More Background On Communications-News.com

MPT4U.com was the online home of Machine Perception Technologies, Inc. (MPT), a San Diego technology company that emerged from research at the University of California, San Diego during an important period in the development of computer vision, machine learning and automated facial-expression analysis.

The company was founded around 2009 to commercialize technology capable of detecting and quantitatively analyzing human facial expressions and other forms of behavior. Although Machine Perception Technologies existed only during a relatively brief period under that name, the research and intellectual property associated with the company became part of a much larger story. MPT evolved into Emotient, an influential emotion-recognition startup that was eventually acquired by Apple in 2016.

That progression makes MPT4U.com more historically significant than its relatively short life as a commercial website might suggest. It documents an early attempt to take sophisticated academic research in machine perception and turn it into practical software for marketing, security, education, human-computer interaction and other fields.

The original MPT4U.com presented applications including smile analysis, fatigue detection and facial-expression analysis. It also promoted software related to the Computer Expression Recognition Toolbox, or CERT, technology developed by researchers associated with UC San Diego's Machine Perception Laboratory.

Today, MPT4U.com is best understood as a historical record rather than the website of an operating technology company. The reconstructed site's own explanation describes it as historical documentation of the now-closed Machine Perception Technologies.

Origins at UC San Diego

Machine Perception Technologies did not begin as a conventional software startup built around an entirely new commercial idea. Its technological foundation came from years of research at UC San Diego.

The university's Machine Perception Laboratory investigated how computers could recognize and respond to the kinds of signals humans routinely use when communicating: facial expressions, gestures, body movements, speech and other behavioral cues.

UC San Diego's Institute for Neural Computation describes the laboratory's goals as developing computer systems capable of recognizing and reacting to natural speech commands, facial expressions, gestures and body motions; using computational systems to investigate problems similar to those faced by the human brain; and studying the statistical structure of natural audiovisual signals.

This was an ambitious research agenda. Rather than simply teaching computers to identify who appeared in a photograph, the researchers wanted machines to understand something about what people were doing and expressing.

That distinction is important to understanding MPT4U.com. Traditional facial recognition generally asks a question such as, "Who is this person?" Machine perception and expression recognition ask different questions: Is the person smiling? Are they frustrated? Are they paying attention? How strongly is a particular facial muscle movement occurring? Is their behavior changing over time?

UC San Diego's 2010 Technology Transfer Office report described MPT as using proprietary video-analysis algorithms to translate human expression into machine-readable information. The company was founded with licensed technology developed in UC San Diego's Machine Perception Laboratory.

The People Behind Machine Perception Technologies

The central figure in the company's development was Javier R. Movellan, a researcher at UC San Diego's Institute for Neural Computation and founder of the Machine Perception Laboratory.

Movellan's work crossed several fields, including machine learning, computer vision, behavioral analysis and human-machine interaction.

Other researchers closely associated with the technology included Marian Stewart Bartlett, Gwen Littlewort Ford, Ian Fasel, Nicholas Butko and others who participated in the Machine Perception Laboratory's research program.

A UC San Diego profile of Machine Perception Technologies listed its founders and leadership as including Movellan as president, Marian Stewart Bartlett as vice president of research and Stanley Kim as chief operating officer. Gwen Littlewort Ford was listed as a senior researcher, with Ian Fasel and Nicholas Butko also part of the technical team.

Paul Ekman, the psychologist whose research on facial expressions and the Facial Action Coding System became enormously influential in emotion research, was associated with MPT through its advisory structure.

The combination was notable: psychologists and behavioral researchers provided ways of describing facial behavior, while computer scientists and machine-learning researchers attempted to automate the measurement process.

Where Machine Perception Technologies Was Located

Machine Perception Technologies was headquartered in San Diego, California, closely linking the company geographically as well as intellectually to UC San Diego.

A 2012 UC San Diego profile listed MPT at 3830 Valley Centre Drive in San Diego, while later trademark records placed Machine Perception Technologies at 6440 Lusk Boulevard in San Diego.

The company's location put it near UC San Diego and the research ecosystem of the La Jolla and Torrey Pines area, one of Southern California's major concentrations of scientific, biotechnology, telecommunications and technology research.

That proximity mattered. MPT was essentially part of the commercialization pipeline surrounding research developed at the university.

UC San Diego's Technology Transfer Office specifically featured Machine Perception Technologies as an example of university technology moving into commercial development.

What MPT4U.com Offered

MPT4U.com was not designed as a general consumer technology portal. Its content reflected a specialized company attempting to commercialize machine-perception technology.

Among the applications presented on the site were smile analysis, fatigue detection and facial-expression analysis.

Smile analysis represented one of the easiest ways for ordinary users to understand the concept. A computer could examine a face and mathematically analyze characteristics associated with a smile rather than merely recording an image.

MPT also explored driver-fatigue detection. This was part of a wider research effort into determining whether measurable behavioral signals could predict declining attention or an increased risk of an accident.

Facial-expression analysis was the broader and potentially more important category. Instead of detecting only a smile, automated systems could attempt to identify multiple facial movements and quantify their intensity.

The site eventually announced a consumer-oriented Android application called SMILE. The application detected smiles, ranked them and placed results on a world map. The reconstructed MPT4U material records a price of $0.99 through Google's Android marketplace, although no original MPT products are currently available.

This was an interesting departure from MPT's research and enterprise orientation. SMILE offered a simple, accessible demonstration of the underlying concept: software could look at a human face and produce data about an expression.

CERT and Automated Facial Action Coding

One of the most significant technologies associated with the MPT story was the Computer Expression Recognition Toolbox, better known as CERT.

CERT was developed by researchers including Gwen Littlewort, Jacob Whitehill, Tingfan Wu, Ian Fasel, Mark Frank, Javier Movellan and Marian Bartlett.

The system attempted to automate aspects of the Facial Action Coding System, or FACS.

FACS breaks facial behavior into individual observable movements known as action units. Rather than simply labeling a face "happy" or "sad," an FACS-based approach can describe specific muscular movements.

Automating this process was important because manual FACS coding requires specialized training and can be extremely labor intensive.

CERT represented a way of letting computers perform facial measurements automatically and in real time.

At the 2008 Neural Information Processing Systems conference, researchers from UC San Diego demonstrated machine-perception technologies for human-machine interaction. Their demonstration included CERT, real-time analysis of auditory information, visual-attention technology and facial-expression analysis for automated tutoring.

Research documentation described CERT as capable of automatically measuring the intensity of numerous facial actions and recognizing several prototypical facial expressions.

Academic distribution of CERT was handled by Machine Perception Technologies under the name AFECT, or Automatic Facial Expression Coding Tool, with academic use offered free of charge.

This provides direct evidence that MPT4U.com was not merely a promotional site. It functioned as a commercial and distribution point connecting academic research with outside researchers and potential users.

From Facial Recognition to Facial Expression Recognition

One source of confusion surrounding MPT is the broad use of the phrase "facial recognition."

The company's work was more specifically concerned with facial-expression recognition and behavioral analysis.

Identity recognition attempts to determine whether a face belongs to a particular person. Expression analysis instead examines visible changes in a person's face.

That difference opened a much wider range of applications.

A camera in a retail environment, for example, might not need to know a shopper's name. A company might instead want aggregated information about whether shoppers appeared interested, surprised or disengaged when viewing a product.

Similarly, educational software might not care about a student's identity but could potentially respond differently if the student appeared confused or disengaged.

This concept — computers adapting to human behavior — was central to the broader machine-perception vision.

Potential Markets for MPT Technology

Machine Perception Technologies envisioned uses extending far beyond a single software application.

UC San Diego described potential MPT systems for marketing, education and security. The company's adaptive technology was intended to learn from and recognize quantified human behavior rather than simply identify people.

Marketing was an obvious commercial market.

Traditional market research depends heavily on questionnaires, focus groups and what consumers say about their reactions. Automated expression analysis promised another source of information: observable behavior while a person actually watched an advertisement, examined packaging or interacted with a product.

Procter & Gamble and Intel were among major companies associated with MPT-related work in contemporary UC San Diego descriptions. The University at Buffalo also identified Machine Perception Technologies and Procter & Gamble as industrial partners connected with automated facial-action-coding technology.

Education offered another intriguing application.

Research demonstrated systems in which facial information could potentially estimate whether students were finding material difficult and adjust educational content accordingly. This anticipated today's broader interest in adaptive and personalized learning technologies.

Security was another potential market, although it was also one of the most controversial.

Automated behavioral analysis raised the possibility that computer systems could identify expressions or patterns believed to indicate stress, unusual behavior or potential threats. Such applications attracted government interest but also raised substantial questions about reliability, privacy and the danger of interpreting ambiguous facial behavior as evidence of intent.

MPT, Paul Ekman and the Popular Culture of Facial Expressions

The timing of MPT's emergence helped make its work especially interesting.

Public awareness of facial-expression analysis increased substantially during the late 2000s, partly through popular culture.

The television series Lie to Me, which premiered in 2009, featured Tim Roth as Dr. Cal Lightman, a character inspired in part by psychologist Paul Ekman's work on facial expressions and deception.

MPT's archived material discussed Ekman, FACS and possible applications of automated expression analysis in areas ranging from security to consumer research.

UC San Diego even used Lie to Me as a recognizable reference when describing Machine Perception Technologies. Its Technology Transfer Office asked what the television series, Sony's Smile Shutter camera technology and Department of Homeland Security airport-security efforts had in common, answering that each was connected in some way with technology or researchers associated with MPT.

This helped place a technically complex startup within a cultural moment when the idea of "reading faces" was attracting broad public attention.

Sony's Smile Shutter and Consumer Technology

One of the most visible examples of the underlying UC San Diego research reaching consumers was Sony's Smile Shutter technology.

UC San Diego later reported that the Machine Perception Laboratory developed an algorithm that became a centerpiece of Sony's Smile Shutter feature.

Smile Shutter allowed compatible cameras to detect when someone smiled and automatically take a photograph.

Today that may sound ordinary. Cameras and smartphones routinely identify faces and scenes automatically. At the time, however, this represented an important change in the relationship between people and cameras.

The camera was no longer simply capturing whatever appeared in front of the lens. It was interpreting the visual scene and making a decision based on human behavior.

That is precisely the type of human-machine interaction the Machine Perception Laboratory and MPT were attempting to advance.

The Transition from MPT to Emotient

Perhaps the most important part of the MPT4U.com story occurred after the original MPT identity began disappearing.

Patent records provide unusually strong evidence for the corporate transition.

A patent concerning anonymization of facial expressions records an assignment to Machine Perception Technologies in May 2013 and then a November 2013 "change of name" from Machine Perception Technologies Inc. to Emotient, Inc.

Other intellectual-property records show the same connection. A Canadian trademark application filed in 2013 lists Machine Perception Technologies as the applicant for the trademark EMOTIENT for facial-expression-recognition and analysis software.

In other words, Emotient should not be viewed merely as an unrelated later company pursuing similar technology. Public intellectual-property records establish a direct corporate connection between the MPT and Emotient names.

This also explains why the public history of MPT seems to end relatively abruptly. The technology and team did not simply disappear.

They moved forward under a new identity.

Emotient Expanded the Commercial Vision

Under the Emotient name, the technology received considerably greater publicity and investment attention.

UC San Diego described Emotient in 2014 as a startup built around technology from Movellan's Machine Perception Laboratory. Movellan and Marian Bartlett were credited with pioneering automated facial coding using computer vision and machine learning.

The company's applications expanded across retail, healthcare and entertainment.

Retailers could potentially analyze customer responses. Healthcare applications could help interpret patient expressions, including pain. Entertainment and gaming systems could theoretically adapt content according to a user's responses.

Emotient increasingly focused on emotion detection and sentiment analysis — concepts that are now frequently grouped under labels such as "emotion AI" or "affective computing."

The company reportedly grew to more than 50 employees by the end of 2015.

That growth provides useful perspective on MPT4U.com's significance. The original MPT operation was small. A modern company database estimates its workforce at only 2–10 employees and records a $2 million seed financing round in 2012. But the technology platform and research team eventually supported a much larger commercial operation.

Apple's Acquisition

The story reached a major milestone in January 2016 when Apple acquired Emotient for an undisclosed amount.

The acquisition generated extensive technology and mainstream press coverage.

UC San Diego reported that Emotient's co-founders Javier Movellan, Marian Stewart Bartlett and Gwen Littlewort left the university to join Apple in Cupertino, along with several former UC San Diego students working at Emotient.

Bloomberg, WIRED, KPBS, InformationWeek, MacRumors and other outlets covered the acquisition.

Apple did not publicly disclose what it intended to do with Emotient's technology.

Nevertheless, the acquisition demonstrated the commercial importance of the field MPT had entered years earlier. What had once been specialized research into automated facial-action coding had become strategically valuable technology to one of the world's largest consumer-technology companies.

Patents and Intellectual Property

The MPT and Emotient story is also documented through a substantial trail of patents and patent applications.

One example concerns automated facial-action coding technology associated with Movellan, Fasel, Littlewort-Ford, Bartlett and Mark Frank.

Another Machine Perception Technologies patent application dealt with operating a machine-learning environment. Additional filings addressed expression recognition, head-pose-invariant analysis, facial-expression anonymization and methods for gathering training data.

The evolution of ownership recorded in patent databases is particularly revealing. Some inventions originated with university researchers or UC San Diego. Others were assigned to Machine Perception Technologies, then Emotient, and eventually Apple.

These records provide a technological paper trail linking academic machine-perception research to MPT and onward to Emotient's commercial portfolio.

Popularity and Public Visibility

MPT4U.com itself never appears to have become a high-traffic mainstream consumer destination.

That was probably never its principal purpose.

Its likely audience consisted primarily of researchers, potential business customers, technology professionals, government organizations, investors and people interested in computer vision and behavioral analysis.

Its importance therefore cannot reasonably be judged by ordinary consumer website popularity.

The people and technology behind it had a much larger reach than the MPT4U brand itself.

Research associated with the Machine Perception Laboratory appeared at major academic conferences such as NeurIPS. CERT circulated among researchers. Sony incorporated related smile-detection research into consumer cameras. Government and commercial organizations investigated possible applications. Emotient later received extensive media coverage, culminating in Apple's acquisition.

MPT4U.com was one relatively small public window into that much larger ecosystem.

Awards, Recognition and Academic Reputation

There is little evidence that MPT4U.com itself won major website or design awards, and it would be misleading to attribute later Emotient awards directly to the MPT4U website.

The more meaningful recognition came through academic publication, university technology-transfer programs and commercial adoption.

UC San Diego repeatedly highlighted the Machine Perception Laboratory and its commercialization efforts. Researchers associated with MPT published extensively in computer vision, machine learning, affective computing and behavioral analysis.

The automated facial-action-coding work also had intellectual-property recognition through university licensing and patent development.

In this context, scientific credibility and successful technology transfer mattered much more than conventional website awards.

Reviews and Criticism

Traditional customer reviews of MPT4U.com or Machine Perception Technologies are scarce.

Again, this reflects the company's specialized business model. MPT was not primarily selling mass-market products through a large consumer storefront.

Evaluation instead occurred through academic research, demonstrations, enterprise relationships and discussion of the broader technology.

Automated emotion recognition has always attracted criticism alongside enthusiasm.

One concern is accuracy. Facial movements do not necessarily have a simple one-to-one relationship with internal emotional states. Context, culture, personality and circumstances matter.

Privacy presents another problem. Systems capable of continuously analyzing faces can potentially collect information about people who never explicitly agreed to participate.

Security applications raise even more serious issues if uncertain behavioral predictions are used to make consequential decisions about individuals.

MPT's historical material itself acknowledged critics who feared intrusive surveillance, "mind reading," questionable science and high costs.

Those concerns have become more rather than less relevant as artificial intelligence and computer vision have grown more powerful.

Cultural and Social Significance

MPT4U.com captures an interesting transitional period in computing.

For decades, people adapted themselves to computers. Users learned commands, keyboards, menus, programming languages and rigid interfaces.

Machine perception proposed something different: computers could learn to adapt to people.

A system might see that someone was smiling. Educational software might recognize confusion. A vehicle might detect fatigue. A robot might respond to facial gestures. Advertising researchers might measure audience reactions without relying entirely on surveys.

This idea is now embedded in the broader development of artificial intelligence.

Modern smartphones organize photographs using computer vision. Vehicles monitor drivers. Video platforms analyze images automatically. Customer-service systems attempt to infer sentiment. Robots increasingly combine cameras, microphones and machine learning to respond to their surroundings.

MPT belonged to an early generation of companies trying to commercialize this transition.

At the same time, its history illustrates why technological capability and social acceptability cannot be separated. A computer capable of reading behavioral signals may make an interface more intuitive, but the same capability can create surveillance and privacy risks.

That tension remains central to today's debates over AI.

What Happened to MPT4U.com?

The original MPT4U.com eventually ceased operating as the active website of Machine Perception Technologies.

The historical material preserved on the later site explains that the domain became available after the original company's presence ended and was subsequently acquired for the purpose of reconstructing and documenting the company's history. It explicitly warns visitors that the company is closed and its products are unavailable.

That makes the present site fundamentally different from the original.

Visitors should not interpret MPT4U.com as a current vendor of expression-recognition technology, nor should historical product descriptions be interpreted as current offers.

Its greatest value today is archival.

The Lasting Importance of MPT4U.com

MPT4U.com represents a small but revealing chapter in the history of artificial intelligence.

Machine Perception Technologies began by attempting to commercialize research that taught computers to quantify and respond to human facial behavior. The company drew upon technology developed at UC San Diego's Machine Perception Laboratory, worked with researchers who were prominent in automated facial coding, and explored applications spanning marketing, education, security and human-computer interaction.

Its technology was associated with CERT and AFECT, smile detection, facial-action analysis and experimental systems capable of interpreting human behavior.

The corporate identity then evolved. Intellectual-property records show Machine Perception Technologies becoming Emotient, while UC San Diego records demonstrate how many of the same researchers and technologies moved into the new company.

Emotient took the commercial idea substantially further, transforming automated facial analysis into a platform for emotion detection and sentiment analysis. Its technology attracted customers, investors and widespread press coverage.

Then Apple acquired the company.

Viewed from that perspective, MPT4U.com was not simply the website of an obscure San Diego startup that disappeared after a few years. It was an early digital home for technology and people who participated in the development of a field that has since become an important part of modern artificial intelligence.

Its history also provides a useful reminder that today's seemingly commonplace AI capabilities emerged from decades of research into difficult questions: How can computers recognize human behavior? Can facial movement be translated into reliable numerical information? Can machines respond intelligently to human emotion? And where should society draw the line when computers become capable of observing and interpreting people?

Those questions remain unresolved.

That is what gives MPT4U.com continuing historical relevance. It preserves evidence of a period when emotion-aware computing was moving from the research laboratory into commercial products — and when the possibilities and problems that surround machine perception today were only beginning to become widely visible.

MPT4U.com