# Processing with AI
## Exploration : IA & Ethics

Name:
> El Ammari Mohamed
>
Subject:
> Detect student presence in class using Face Recognition

>[TOC]

## Design brief

### Biais

If we don't source our dataset with enough rigor, the following biais might appear:

>1. Ethnicity : Missrecognition of the student due to a dataset that include more white people than minorities
>2. Gender: It appears that not enough rigour would lead to false detection such as missgender identification
>3. Quality: if the model doesn't include dataset of images with high quality it may lead to trouble in detection.

We will ensure that our model is not biaised by:

>1. Sourcing our data from already existing and accessible dataset if possible such as IBM, Microsoft, Face++, Kairos... Otherwise our dataset should include both online collection of pictures of students from all partner universities, and also think about scanning old pictures of students that might be used to train and make the model more accurate.
>2. Making sure our data take into account both genders, and ethinicity in equal data collection so that the model won't be biased.
>3. The model should also be train to detect more angles in order to have a high accuracy in avoiding false identification of the students

### Overfitting

We will make sure our model does not overfit by
> Separating the dataset of the different pictures of students in two parts, a training dataset and a validation dataset.
> Our model will be trained using the training one, then we will check the accuracy on the validation one.

### Misuse

>We have to remind ourselves that our application could be misused by any employee within the school who would like to track a student and know more when was he or she identified, and where mainly, consequently is any sexual predator is within the school staff it could be a potential danger in misusing the application.

### Data leakage

>In a catastrophic scenario, where all of our training dataset were stolen or recovered from our model, the risk would be that all our students' information "biometric data" could be used to:
>-Provide false ID card
>-Deploy the presence of sensitive student in the school
>-Delete or add the presence of new users without consent
>-The breach could also lead to access the database of students with all the necesssary informaiton about also their parents too.

### Hacking

> If someone found a way to "cheat" our model and make do it make any prediction that it want instead of the real one, the risk would be that all the attendance recorded would be misleading for the school since the data matching would be false which will make the school weak in term of reliability, and parents would not take the risk that their data could be hacked or be cheated by any unpleasant process.