Page 70 - Raspberry Pi as a Foundation for Boosting Computer and Technology Literacy

5 Raspberry Pi usage
across industries. Thus, numerous tools have been created to simplify
and speed up the process of developing and implementing machine
learning solutions. When it comes to Raspberry Pi devices, there are
two common approaches for delving into machine learning. These ap-
proaches involve either utilizing external Python modules or making
use of Mathematica’s built-in functions.

Python libraries
In order to take advantage of Python’s machine learning libraries,
one would most probably need to install such libraries as Seaborn
(Waskom 2021), Pandas (McKinney 2010), Numpy (Harris et al. 2020),
and Matplotlib (Hunter 2007). These libraries are not only common
dependencies but also provide useful methods for data processing,
manipulation, and visualization.
The subsequent steps depend completely on the particular machine
learning model. For instance, if one intends to generally explore and
experiment with a wide range of machine learning algorithms, install-
ing Scikit-learn (Pedregosa et al. 2011) could be a viable option. Simi-
larly, the OpenCV (Bradski 2000) library serves as a versatile solution
for various computer vision projects (Monk 2022, 225–238). In addi-
tion, the Pillow ( https://python-pillow.org) library offers a range of
general image processing methods that are useful for the pre-process-
ing and augmentation of the image data set.
On the other hand, such libraries as Keras ( https://keras.io) and
TensorFlow (Abadi et al. 2016) provide a comprehensive solution for
deep-learning tasks. TensorFlow offers a range of pre-trained models
that enable users to experiment with object and whistle detection, as
well as sound identification (Monk 2022, 240–48). Conversely, Keras
is a high-level framework that uses TensorFlow as a backend. Thus, it
provides modules and methods designed for developing and training
neural networks, saving and importing models, applying various acti-
vation functions, among others (Norris 2019, 324–325). However, one
would need to switch to the 64-bit version of Raspberry Pi OS (Rasp-
berry Pi Foundation, n.d.-a), as all three libraries are incompatible
with 32-bit systems.

Wolfram language
Mathematica includes an array of pivotal machine learning algorithms,
allowing its users to perform linear regression, logistic regression, and

70