Gillespie's Algorithm [draft] and Face Expression Recognition

Tarleton Gillespie addresses the issue of algorithms having become a mystified 'talisman' in the contemporary public imagination, or simply that the definition of one is not adequately understood.

"For “algorithm,” there is a sense that the technical communities, the social scientists, and the broader public are using the word in different ways. For software engineers, algorithms are often quite simple things; for the broader public they name something unattainably complex."

What is most often referred to in this way are the secret black-box algorithms deployed by social media companies to curate and moderate content feeds or display targeted advertisements. It is in the companies' interests to keep their methods unattainable, for a plethora of reasons: one that is often brought up is the fight against (what, for example, Youtube calls) 'bad actors' seeking to cheat the system.

"An algorithm is a recipe composed in programmable steps"

Engineers choose between them based on values such as how quickly they return the result, the load they impose on the system’s available memory, perhaps their computational elegance. The embedded values that make a sociological difference are probably more about the problem being solved, the way it has been modeled, the goal chosen, and the way that goal has been operationalized (Reider).

The most common problem in algorithm design is that the new data turns out not to match the training data in some consequential way.

For algorithm designers, the algorithm is the conceptual sequence of steps, which should be expressible in any computer language, or in human or logical language. 

questions of value are very much bracketed in the early decisions about how to operationalize a social activity into a model and into the miniscule, mathematical moments of assigning scores and tuning thresholds.

Asserts that there's a high cultural value(?) for statistical conclusions that are objective in a way humans can't be- and then acknowledges where this is changing as information companies (big term) use algorithms as a sort of scapegoat for their mistakes?

Questions of culpability/anonymity of algorithm programmers + devs... hypothetical identification of workers could be bad...


Face expression recognition (FER) algorithms can be analysed well along the lines of Gillespie's article. There is a lot to be said about the relationship between the human at the preprocessing stage, the computer, in processing the dataset, and the human receiving the results/output. Facial expressions are a highly sensitive and emotive subject for humans, being our most pertinent form of nonverbal communication. The idea that a computer might be able to determine our true emotions by processing images is a good exemplary case with which to examine the traits of algorithm authorship.

examples of databases

Carnegie Mellon University

Experiments are performed on FER by using various databases likely Japanese Female Facial Expressions (JAFFE, 2017), Cohn – Kanade (CK, 2017), Extended Cohn – Kanade (CK+), MMI (MMI, 2017), Multimedia Understanding Group (MUG, 2017), Taiwanese Facial Expression Image Database (TFEID, 2017), Yale (Yale, 2017), AR face database (AR, 2018), Real-time database (Zhao and Pietikäinen, 2009), Own database (Siddiqi et al., 2015) and Karolinska Directed Emotional Faces (KDEF, 2018). In most of the experiments, JAFFE database is used. JAFFE holds ten Japanese female’s expressions with seven facial expressions and totally 213 images. Each image in JAFFE database contains 256 x 256 pixel resolution.[2]

The different databases come with their own unique sets of expression types

Angry, Fearful, Disgusted, Sad, Happy, Surprised, Neutral

Neutral, anger, contempt, disgust, fear, happiness, sadness, surprise

Happy, normal, sad, sleepy, surprised, wink

Neutral, smile, anger, scream [2] (+more)

Face recognition is a relevant subject in pattern recognition, neural networks, computer graphics, image processing and psychology [125]. In fact, the earliest works on this subject were made in the 1950’s in psychology [21]. They came attached to other issues like face expression, interpretation of emotion or perception of gestures. Engineering started to show interest in face recognition in the 1960’s. One of the first researches on this subject was Woodrow W. Bledsoe. In 1960, Bledsoe, along other researches, started Panoramic Research, Inc., in Palo Alto, California. The majority of the work done by this company involved AI-related contracts from the U.S. Department of Defense and various intelligence agencies [1]

The method proposed in this paper is ideal for classifying unknown facial expressions of person dependent, reaching over 95%, but the classification effect on unknown facial expressions of person independent is about 74%. This is mainly because: (1) for different face shapes, the difference in contour is relatively large, and the change between different expressions is sometimes very weak, thus causing serious interference to facial expression recognition; (2) the different faces participating in the classifier training are less. 3) As the downsampling factor increases, the recognition rate changes are not very obvious, but the calculation load is greatly reduced. [3, p.67]

[1] http://www.ehu.eus/ccwintco/uploads/e/eb/PFC-IonMarques.pdf

[2] https://www.sciencedirect.com/science/article/pii/S1319157818303379#f0010

[3] http://www.ijsps.com/uploadfile/2019/0712/20190712023146977.pdf