=Paper=
{{Paper
|id=Vol-3762/578
|storemode=property
|title=AI for Sustainability: Activities of the CINI-AIIS Lab at University of Naples Federico II
|pdfUrl=https://ceur-ws.org/Vol-3762/578.pdf
|volume=Vol-3762
|authors=Flora Amato,Giovanni Giacco,Lidia Marassi,Stefano Marrone,Antonio Elia Pascarella,Carlo Sansone
|dblpUrl=https://dblp.org/rec/conf/ital-ia/AmatoGM0PS24
}}
==AI for Sustainability: Activities of the CINI-AIIS Lab at University of Naples Federico II==
https://ceur-ws.org/Vol-3762/578.pdf
AI for Sustainability: Activities of the CINI-AIIS Lab at
University of Naples Federico II
Flora Amato1 , Giovanni Giacco1,2 , Lidia Marassi1 , Stefano Marrone1,* , Zahida Mashaallah1 ,
Antonio Elia Pascarella1 and Carlo Sansone1
1
Department of Electrical Engineering and Information Technology (DIETI), University of Naples Federico II, Via Claudio 21, 80125 Naples, Italy
2
Latitudo 40, 80146 Naples, Italy
Abstract
Sustainability is pivotal to global development, aligning closely with the United Nations’ goals for a sustainable future. This
paper introduces and discusses the perspectives and initiatives undertaken in these regards by the CINI AI-IS (the Italian
National Consortium for Informatics, Artificial Intelligence and Intelligent Systems) Lab at the University of Naples Federico II.
We will first introduce the DroughtScope project, currently on board the Kanyni Australian satellite to exploit hyperspectral
data to detect early water stress in crops and optimize water resource management. We will then describe the PIVA project,
addressing the challenge of missing data in complex systems, which occurs frequently in environmental domains, using
Physics-Informed Variational Auto-Encoders to prevent model collapse. Additionally, the impact of Agriculture 4.0 on farmer
health and workplace safety is discussed, examining the challenges and opportunities presented by advanced technologies.
Finally, the paper considers the environmental and ethical implications of AI’s carbon footprint, emphasizing the need for a
balanced approach to technological advancement and environmental accountability.
Keywords
Synthetic data, Carbon footprint, Ethics, Human-Centred AI
1. Introduction ronmental changes, and facilitating more informed deci-
sions about natural resource management. AI’s capability
Advancements and integrative applications of artificial to process and analyze vast amounts of environmental
intelligence (AI) in agritech and environmental sustain- data enhances our ability to respond to climate change,
ability are becoming increasingly important as the global manage natural disasters, and protect biodiversity. How-
community seeks innovative solutions to pressing envi- ever, the application of AI in sustainability also raises
ronmental challenges. The integration of AI technologies important ethical and practical challenges, including the
in agricultural and environmental contexts promises to risk of increased energy consumption, potential biases in
enhance efficiency, reduce resource waste, and improve decision-making processes, and the implications for em-
decision-making processes, aligning with several of the ployment in traditional farming and environmental con-
United Nations Sustainable Development Goals (SDGs), servation roles. It is essential to address these challenges
such as responsible consumption and production (Goal by developing AI solutions that are not only effective but
12) and climate action (Goal 13) [1]. Artificial intelli- also equitable and inclusive.
gence offers a transformative potential for the agricul- In this paper, we will thus introduce and discuss the
tural sector by optimizing resource use and maximizing perspectives and initiatives undertaken on responsible
output, thereby addressing food security and economic and reliable AI by the CINI AI-IS (the Italian National
sustainability. For instance, AI-driven systems can pre- Consortium for Informatics, Artificial Intelligence and
dict crop yields, monitor crop health through real-time Intelligent Systems) Lab at the University of Naples Fed-
data, and provide precise inputs regarding irrigation and erico II, specifically focusing on the activities involving
fertilization, significantly reducing unnecessary resource the members of the PICUS Lab1 as part of the AI-IS Node.
expenditure and environmental impact. To this aim, Section 2 describes the DroughtScope project,
In environmental management, AI technologies are a finalist in the ESA’s OrbitalAI IMAGIN-e competition2 ,
critical in monitoring ecosystem health, predicting envi- using hyperspectral data to optimize water resource man-
agement through early detection of water stress in crops
Ital-IA 2024: 4th National Conference on Artificial Intelligence, orga-
nized by CINI, May 29-30, 2024, Naples, Italy
and the generation of alerts for risk areas. Section 3
*
Corresponding author. discusses the use of AI for cows’ mastitis detection, an
$ stefano.marrone@unina.it (S. Marrone) inflammatory condition of the udder causing critical is-
0000-0002-5128-5558 (F. Amato); 0000-0001-7143-1244 sues for dairy milk and animal health. Section 4 describes
(G. Giacco); 0009-0006-8134-5466 (L. Marassi); 0000-0001-6852-0377 the PIVA project, focusing on missing data imputation
(S. Marrone); 0000-0002-1079-7741 (A. E. Pascarella);
0000-0002-8176-6950 (C. Sansone) 1
https://picuslab.dieti.unina.it/
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License 2
Attribution 4.0 International (CC BY 4.0). https://platform.ai4eo.eu/orbitalai-imagin-e
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
Figure 1: Raster of prediction and ground truth of the DroughtScope project.
in complex systems by means of Physics-Informed Varia- This project is currently on board the Kanyni Australian
tional Auto-Encoders. Section 5 focuses on the challenges satellite. Some preliminary results are reported in Table
and opportunities for improving farmers’ well-being and 1, while Figure 1 shows some predictions made by the
productivity in the era of Agriculture 4.0. Finally, Sec- model.
tion 6 analyses the AI environmental impact, particularly
focusing on the carbon footprint of large-scale AI models. accuracy recall precision f1 model size
0.8 0.7 0.7 0.7 90 MB
2. DroughtScope project
Table 1
The DroughtScope project, which appears among the fi- Performance of the DroughtScope project.
nalists in the OrbitalAI IMAGIN-e competition organized
by the European Space Agency (ESA), focuses on identi-
fying early water stress conditions in crops to optimize
water resource management. Utilizing hyperspectral data 3. Data Analysis over Mastitis
from IMAGIN-e, DroughtScope estimates evapotranspira-
tion (ET) at a plot scale, enabling the creation of synthetic
Detection
indicators for early detection of water stress. The project Mastitis is a critical issue for dairy milk and animal health.
employs a multi-task deep-learning network to produce It is an inflammatory condition of the udder which ef-
a real-time crop/no crop classification map and the Evap- fects economically reduces milk yield. Several method-
orative Stress Index (ESI) product. The project uses open ologies, biological and new AI-based mastitis detection
data from the ESA WorldCover dataset for ground truth use machine learning algorithms and artificial intelli-
in crop mapping. Additionally, ECOSTRESS data, with a gence to analyse data from multiple sources, such as milk
spatial resolution of approximately 70 meters, are used production records, udder health parameters and sen-
for ESI and are rescaled to 45 meters to align with the sor readings, to identify patterns that indicate the onset
input hyperspectral data. The DroughtScope’s architec- of mastitis in dairy cows. By continuously monitoring
ture is designed to economize memory usage by sharing and analysing these data points, AI systems can provide
a single encoder across multiple tasks and leveraging early and accurate detection of mastitis, enabling timely
a feature-based Knowledge Distillation (KD) technique. intervention and optimisation of dairy herd health and
Table 2
An overview of recent works on mastitis detection.
Reference No. Specific Problem Techniques Problem Solution Models
[2] Buffalo Mastitis Detection disease The uddar size feature of buffalo is fused with temperature Automatic diagnosis of early stage mastitis in buf- Neural Network
feature. To automatically detect uddar and eye of buffalo falo.Optimal AI-baed management of commercial farms YOLO7,CenterNet,SVM
need neural network model YOLO7 and extract correspond-
ing temperature and create temperature feature vector. Cen-
terNet check the size of uddar and create size feature vec-
tor.Fused temperature feature vector and size feature vec-
tor.SVM(support vector machine) measure the degree of
mastitis.
[4] Buffalo Mastitis Detection disease Ultrsonography images of buffalo for training deep learnig Buffalo Mastitis Detection combination of deeplearnig model Ultrasound uddar
model, EfficientNet,Polyloss,Convolutioanl block attention and ultrasound images images+Efficientb3 net-
module, Somatic cell count work+CBAM+Somatic
cell count+Polyloss
generete a model for
mastitis detection
[5] Buffalo mastitis detection disease Thermal infrared mastitis detection technology automati- The cow thermal infrared acquisition system, Accurate de- CLE-UNet Network
cally segments key parts of the cow’s eyes and udder in ther- tection of cow mastitis in large-scale dairy farms Model
mal infrared image segmentation technology. CLE−UNet
(Centroid Loss Ellipticization UNet) semantic segmentation
algorithm, ECA (efficient channel attention), Lovasz softmax
loss function, FLIR tools.
[3] Mastitis detection based on udder The datasets used include data collected from the udder by The use of machine learning classifiers to detect cow diseases Classification Model in-
characteristics and temperature. four flex sensors and one temperature sensor. Machine learn- from images and associated metadata cludes RF,SVM,KNN,NB
ing classifier training includes Decision Tree (DT), Naive Bye and DT
(NB), Support Vector Machine (SVM), K−Nearest Neighbour
(K−NN) and Random Forest Algorithm (RF)
[6] Machine learning analysis to pre- Prediction model developed using four different machine Machine learning methods applied to improve subclinical Subclinical mastitis pre-
dict the presence or absence of sub- learning algorithms Generalised linear model, support vector mastitis prediction model diction models include
clinical mastitis in Italian buffaloes machine, random forest and neural network, Support Vector Generalised Linear Model,
Machine to predict high or low somatic cell SVM, an algorithm based
on decision tree(RF), pat-
tern recognition.
productivity. In recent years, researchers have more and missing completely at random (MCAR), missing at ran-
more using AI for detecting mastitis in the early stages of dom (MAR), and missing not at random (MNAR). MCAR
the disease. In 2023, an automatic detection method for occurs when data missingness is entirely independent of
dairy cow mastitis using the fusion of udder temperature observed or unobserved variables. MAR happens when
and size features based on deep learning was proposed the probability of missingness is related to the observed
[2]. The author used the YOLO7 model, centre network data but not the missing data itself. Finally, MNAR arises
and Support Vector Machines in his research, showing when the missingness is related to the unobserved data.
promising performance in the early detection of masti- Over the years, different approaches have been pro-
tis. In the same year, the Cina Agriculture University posed to perform missing data imputation, i.e. the recon-
showed that thermal infrared technology combined with struction of missing pieces of information starting from
a deep CLE-Unet model can significantly improved the available ones. As for several other domains, recently
detection accuracy of mastitis [3]. More recently, deep Deep Learning (DL) solutions are becoming more fre-
learning combined with buffalo udder ultrasound was quent. However, a significant challenge with DL-based
used for the first time to detect mastitis with the aim of systems, such as those based on Variational Auto-Encoder
establishing an accurate, fast and inexpensive method (VA) [7, 8], is their susceptibility to model collapse, where
to detect buffalo mastitis instead of routine laboratory results often converge to median values. As this would
examination [4]. The model is suitable for mastitis detec- make the imputed data statistically not compliant with
tion and can be used in a small dairy farm, but still needs the underlying physical phenomena, it is crucial to de-
more deep learning methods to detect the disease more velop systems robust against this issue.
finely. Besides the reported example, Table 2 reports and The PIVA (Physics Informed Variational Auto-Encoder)
details other approaches recently used for the task. project is a solution specifically designed to prevent
model collapse. To this aim, PIVA incorporates a dual
mechanism approach, utilizing both physical constraints
4. The Piva project and a masking technique. It is structured on a Variational
Auto-Encoder (VAE) architecture where constraints are
In the era of data-driven decision-making, the quality and
integrated into the loss function to guide the network to-
completeness of data play a crucial role across diverse
wards adhering to essential physical and statistical param-
fields, ranging from medical to industrial and environ-
eters. These include entropy conservation, summation
mental applications. In all these cases, missing data is a
constraints across variable groups, control over covari-
frequent problem that can arise for various reasons, in-
ances, and adherence to the Wasserstein distance. The
cluding sensor malfunctions or human errors. Based on
technique of conserving entropy in both generated and
the particular underlying reason, the phenomena of miss-
observed variables has been demonstrated to be particu-
ing data are categorized into three distinct mechanisms:
larly effective in mitigating model collapse. Additionally,
PIVA adopts a novel data masking strategy inspired by cycle of nature, of contributing to food production and
the method used in Bert. Unlike traditional Denoising community well-being, can be shaken when machines
Auto-Encoders that focus on reconstructing data from begin to perform these roles, leading to a sense of loss
noise-altered inputs, PIVA’s strategy concentrates on ac- of identity, which can have negative effects on farmers’
curately predicting masked data while ensuring compli- mental health. Also, the risk of social isolation due to
ance with the constraints imposed on these data points. decreased human interaction can have a significant im-
This approach is pivotal for achieving dependable data pact on mental health, causing feelings of loneliness and
imputation and enhancing the overall robustness of the depression. Although Agriculture 4.0 undoubtedly brings
model. advantages in terms of production and sustainability, it
is essential to recognize and address the potential neg-
ative effects on farmers’ health, particularly those of a
5. Farmer Health in the Era of psychological nature.
Agriculture 4.0: Challenges and To ensure the prudent and efficient development of 4.0
solutions in agriculture, regulations and directives should
Opportunities be implemented to ensure a safe and healthy working
In the context of Industry 4.0, Agriculture 4.0 represents a environment and promote the physical and mental well-
crucial step in the evolution of precision agriculture. Au- being of agricultural workers. At the same time, it is
tomation, the use of drones and sensors, data collection, essential to develop psychological support programs, pro-
and artificial intelligence have enabled a more precise vide resources to address change and promote a culture
and efficient approach to agriculture, promising to in- of mental well-being within agricultural communities to
crease production and reduce waste. Moreover, 4.0 solu- preserve the health and well-being of agricultural work-
tions emerge in response to climate change, contributing ers. The transition to Agriculture 4.0 offers undeniable
to mitigating its negative effects on crop yield, manage- advantages in terms of efficiency and sustainability, but
ment difficulties, and farmer well-being [9]. Undoubtedly, it is essential to consider the ethical implications in terms
the advent of Agriculture 4.0 marks a significant break- of social justice, environmental sustainability, and farmer
through in the agricultural sector, traditionally charac- health. Only through a multidisciplinary, fair, and sus-
terized by a strong reliance on manual outdoor labour, tainable approach, it may be possible to fully realize the
presenting itself as a valuable ally for those who work potential of this technological revolution in agriculture,
the land. For example, thanks to automation and the use ensuring that the benefits are shared fairly and responsi-
of advanced agricultural machinery, farmers can reduce bly by all members of society.
their direct exposure to adverse weather conditions. If
autonomous tractors and drones allow work in the fields 6. The Carbon Footprint of AI:
even in the presence of heavy rains, extreme tempera-
tures, or excessive heat, thus reducing the risk of weather- Ethics and Environmental
related diseases, optimizing agricultural processes allows Accountability
farmers to also plan their activities more intelligently,
avoiding the hottest hours of the day or adverse weather The widespread introduction of artificial intelligence at
conditions [10]. Similarly, in terms of workplace safety, virtually every societal level prompts deep reflection on
a particularly delicate issue in the industrial context, au- the consequences of the massive use of these technolo-
tomation and the use of advanced machinery reduces the gies. This raises fundamental ethical questions about
risk of accidents [11], thereby contributing to preserving how we should regulate and manage these innovations
the health and lives of farmers. to ensure a positive impact on society as a whole. A
However, while these advancements undoubtedly offer highly topical issue is the environmental impact of AI,
benefits in terms of production and sustainability, it is particularly the carbon footprint of learning models. The
essential to recognize and address the potential negative increase in the size of artificial intelligence models, espe-
effects on farmers’ health, particularly those of a psycho- cially those based on deep neural networks (DNNs), con-
logical nature. The gradual replacement of traditional sequently results in higher energy consumption during
tasks of agricultural workers with machines and auto- the training process [12]. This phenomenon is driven by
mated systems, while increasing production and reducing the need for larger models to achieve better performance
physically demanding work, could easily create a sense but raises concerns about the environmental impact due
of alienation for those who have dedicated themselves to increased energy consumption. The fundamental ques-
to agriculture for generations. The sense of personal and tion revolves around striking a balance between AI’s pre-
cultural identity is often deeply connected to agricultural cision goals and the environmental impact resulting from
work. The perception of being an essential part of the such research. Essentially, to what extent is it ethical to
pursue AI research and development focusing solely on our planet and future generations. Furthermore, it is im-
model accuracy if it entails increased energy costs and portant to adopt a proactive approach in defining policies
pollution? It’s a moral trade-off that requires careful and regulations that guide the responsible development
balancing: on one hand, the accuracy of AI models is and use of AI, ensuring that the interests of society as a
crucial for many applications; on the other hand, the whole are adequately represented through a holistic and
rise in energy costs and pollution raises ethical concerns collaborative approach.
about the sustainability of this approach. Furthermore,
complicating this equation is the awareness that much
of AI’s energy costs come from the operational use of Acknowledgments
models [13], highlighting the environmental responsibil-
This work was partially supported by PNRR MUR Project
ity not only of researchers and developers but also of AI
PE0000013-FAIR.
companies, energy providers, and various stakeholders
involved. Identifying and fairly distributing responsi-
bilities among the various actors involved can thus be References
challenging, as they may have conflicting interests and
viewpoints. Developers may focus on innovation and [1] M. Bexell, K. Jönsson, Responsibility and the united
model accuracy, while AI companies may be incentivized nations’ sustainable development goals, in: Forum
to maximize profits, ignoring environmental impacts. Ad- for development studies, volume 44, Taylor & Fran-
ditionally, energy providers may resist transitioning to cis, 2017, pp. 13–29.
more sustainable energy sources for economic reasons. [2] M. C. Q. L. Y. W. X. Z. Y. .GangLiu, Fusion of ud-
These considerations underscore the need for a thor- dar temprature and size features for the automatic
ough reflection on responsible resource management by detection of dairy cow mastitis using deep learn-
society as a whole to mitigate the environmental im- ing, Computers and Electronics in Agriculture 212
pact of AI from the perspective of ethical and sustain- (2023). doi:10.1016/j.compag.2023.108131.
able progress. The consequences of artificial intelligence [3] A. K. M. S. S. Sachdeva, Detection of mastitis disease
do not only concern the technical field. Today, ethics in cow with machine learning classifiers, Research
therefore play a central role in addressing the various Gate 7 (2023) 112–129. doi:10.52865/DQFZ601.
challenges posed by artificial intelligence, and balancing [4] X. J. Y. L. Y. Z. Z. Y. W. Z. P. N. L. Yang,
technological progress with environmental responsibil- A new method to detect buffualo mastitis us-
ity is a particularly delicate moral issue. What seems ing uddar ultrasonography based on deep learn-
desirable and necessary is a coordinated effort by in- ing, MDPI 14 (2024) 707. doi:https://www.mdpi.
dustry, academia, governments, and civil society to find com/2076-2615/14/5/707#.
balanced solutions that take into account both techno- [5] Q. Z. Y. Y. G. L. Y. N. J. Li, Dairy cow mas-
logical progress and sustainability needs [14]. However, titis detection by thermal infrared images based
this approach requires open and transparent dialogue on cle−unet, MDPI 13 (2023) 2211. doi:https:
among all stakeholders, as well as targeted policies and //www.mdpi.com/2076-2615/13/13/2211#.
incentives that promote environmental and social respon- [6] T. B. R. G. A. G. S.Beffani, Exploiting ma-
sibility in technological innovation. For example, inter- chine learning methods with monthly routine milk
national organizations and regulatory authorities can recording data and climate information to predict
collaborate to develop sustainability standards for AI, in- subclinical mastitis in italian mediterranean buf-
cluding environmental criteria to be respected during the faloes, NIH 106 (2023) 1942–1952. doi:10.3168/
development, implementation, and use of AI models (an jds.2022-22292.
example is the International Telecommunication Union, [7] Y. Sun, J. Li, Y. Xu, T. Zhang, X. Wang, Deep learn-
ITU, which has established a working group on envi- ing versus conventional methods for missing data
ronmental issues related to AI, tasked with developing imputation: A review and comparative study, Ex-
recommendations and guidelines to promote sustainable pert Systems with Applications (2023) 120201.
use of the technology) [15]. In an era where technolog- [8] J. Yoon, J. Jordon, M. Schaar, Gain: Missing data
ical innovation is advancing at an unprecedented pace, imputation using generative adversarial nets, in: In-
it is essential to consider the long-term implications for ternational conference on machine learning, PMLR,
the environment and society. The adoption of AI offers 2018, pp. 5689–5698.
enormous benefits in terms of efficiency, automation, and [9] D. C. Rose, R. Wheeler, M. Winter, M. Lobley, C.-A.
performance improvement, but it must be guided by ethi- Chivers, Agriculture 4.0: Making it work for people,
cal values and principles of social equity. This is because production, and the planet, Land use policy 100
the decisions made today regarding the development and (2021) 104933.
implementation of AI will have a significant impact on
[10] V. Goltyapin, I. Golubev, Global trends in the de-
velopment of monitoring systems for mobile agri-
cultural equipment, in: E3S Web of Conferences,
volume 157, EDP Sciences, 2020, p. 01013.
[11] N. Stacey, P. Ellwood, S. Bradbrook, J. Reynolds,
H. Williams, D. Lye, Foresight on new and emerg-
ing occupational safety and health risks associated
with digitalisation by 2025, Luxembourg: European
Agency for Safety and Health at Work (2018).
[12] R. Schwartz, J. Dodge, N. A. Smith, O. Etzioni, Green
ai, Communications of the ACM 63 (2020) 54–63.
[13] D. Patterson, J. Gonzalez, Q. Le, C. Liang, L.-M.
Munguia, D. Rothchild, D. So, M. Texier, J. Dean,
Carbon emissions and large neural network train-
ing, arXiv preprint arXiv:2104.10350 (2021).
[14] G. Tamburrini, The ai carbon footprint and respon-
sibilities of ai scientists, Philosophies 7 (2022) 4.
[15] H.-T. Liao, C.-L. Pan, Y. Zhang, Smart digital plat-
forms for carbon neutral management and services:
Business models based on itu standards for green
digital transformation, Frontiers in Ecology and
Evolution 11 (2023) 1134381.