Skip to content

Commit 7fe9a91

Browse files
iargandaiarganda
authored
Merge pull request #28 from CVPD/hugoblox-import-publications
Hugo Blox Builder - Import latest publications
2 parents 61be6ea + ded597c commit 7fe9a91

4 files changed

Lines changed: 102 additions & 0 deletions

File tree

‎content/publication/abou-2025-agentic/cite.bib‎

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,13 @@
1+
@article{abou2025agentic,
2+
abstract = {Agentic AI represents a transformative shift in artificial intelligence, but its rapid advancement has led to a fragmented understanding, often conflating modern neural systems with outdated symbolic models—a practice known as conceptual retrofitting. This survey cuts through this confusion by introducing a novel dual-paradigm framework that categorizes agentic systems into two distinct lineages: the symbolic/classical (relying on algorithmic planning and persistent state) and the neural/generative (leveraging stochastic generation and prompt-driven orchestration). Through a systematic PRISMA-based review of 90 studies (2018-2025), we provide a comprehensive analysis structured around this framework across three dimensions: (1) the theoretical foundations and architectural principles defining each paradigm; (2) domain-specific implementations in healthcare, finance, and robotics, demonstrating how application constraints dictate paradigm selection; and (3) paradigm-specific ethical and governance challenges, revealing divergent risks and mitigation strategies. Our analysis reveals that the choice of paradigm is strategic: symbolic systems dominate safety-critical domains (e.g., healthcare), while neural systems prevail in adaptive, data-rich environments (e.g., finance). Furthermore, we identify critical research gaps, including a significant deficit in governance models for symbolic systems and a pressing need for hybrid neuro-symbolic architectures. The findings culminate in a strategic roadmap arguing that the future of Agentic AI lies not in the dominance of one paradigm, but in their intentional integration to create systems that are both adaptable and reliable. This work provides the essential conceptual toolkit to guide future research, development, and policy toward robust and trustworthy hybrid intelligent systems.},
3+
author = {Mohamad Abou Ali and Fadi Dornaika and Jinan Charafeddine},
4+
doi = {10.1007/s10462-025-11422-4},
5+
journal = {Artificial Intelligence Review},
6+
number = {1},
7+
pages = {11},
8+
publisher = {Springer},
9+
title = {Agentic AI: a comprehensive survey of architectures, applications, and future directions},
10+
url = {https://doi.org/10.1007/s10462-025-11422-4},
11+
volume = {59},
12+
year = {2025}
13+
}

‎content/publication/abou-2025-agentic/index.md‎

Lines changed: 39 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,39 @@
1+
---
2+
title: 'Agentic AI: a comprehensive survey of architectures, applications, and future
3+
directions'
4+
authors:
5+
- Mohamad Abou Ali
6+
- Fadi Dornaika
7+
- Jinan Charafeddine
8+
date: '2025-01-01'
9+
publishDate: '2026-09-09T07:58:31.731271Z'
10+
publication_types:
11+
- article-journal
12+
publication: '*Artificial Intelligence Review*'
13+
doi: 10.1007/s10462-025-11422-4
14+
abstract: 'Agentic AI represents a transformative shift in artificial intelligence,
15+
but its rapid advancement has led to a fragmented understanding, often conflating
16+
modern neural systems with outdated symbolic models—a practice known as conceptual
17+
retrofitting. This survey cuts through this confusion by introducing a novel dual-paradigm
18+
framework that categorizes agentic systems into two distinct lineages: the symbolic/classical
19+
(relying on algorithmic planning and persistent state) and the neural/generative
20+
(leveraging stochastic generation and prompt-driven orchestration). Through a systematic
21+
PRISMA-based review of 90 studies (2018-2025), we provide a comprehensive analysis
22+
structured around this framework across three dimensions: (1) the theoretical foundations
23+
and architectural principles defining each paradigm; (2) domain-specific implementations
24+
in healthcare, finance, and robotics, demonstrating how application constraints
25+
dictate paradigm selection; and (3) paradigm-specific ethical and governance challenges,
26+
revealing divergent risks and mitigation strategies. Our analysis reveals that the
27+
choice of paradigm is strategic: symbolic systems dominate safety-critical domains
28+
(e.g., healthcare), while neural systems prevail in adaptive, data-rich environments
29+
(e.g., finance). Furthermore, we identify critical research gaps, including a significant
30+
deficit in governance models for symbolic systems and a pressing need for hybrid
31+
neuro-symbolic architectures. The findings culminate in a strategic roadmap arguing
32+
that the future of Agentic AI lies not in the dominance of one paradigm, but in
33+
their intentional integration to create systems that are both adaptable and reliable.
34+
This work provides the essential conceptual toolkit to guide future research, development,
35+
and policy toward robust and trustworthy hybrid intelligent systems.'
36+
links:
37+
- name: URL
38+
url: https://doi.org/10.1007/s10462-025-11422-4
39+
---

‎content/publication/alirezazadeh-2025-cosine/cite.bib‎

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,13 @@
1+
@article{Alirezazadeh2025cosine,
2+
abstract = {Breast cancer claims thousands of lives annually, emphasizing the need for swift and accurate histopathology image classification to expedite diagnoses. Despite the rapid evolution of Convolutional Neural Networks (CNNs), the conventional softmax loss lacks the robustness required to discern intricate features in breast cancer classification from histopathological images. In response, our study introduces the Cosine Additive Angular Margin Loss (CAAM) to address this limitation and achieve both enhanced intra-class cohesion and distinct inter-class boundaries concurrently. By normalizing weight and feature vectors, we eliminate radial variations before imposing angular and cosine margin constraints on the softmax angle space. This process maximizes the decision margin, resulting in a discriminative feature embedding. Extensive experiments conducted on the BreakHis dataset demonstrate that CAAM consistently outperforms existing methods in breast cancer histopathological image classification. Our findings underscore the efficacy of angular margin-based softmax losses in bolstering the performance of advanced CNN models for breast cancer classification.},
3+
address = {New York, NY, USA},
4+
author = {Pendar Alirezazadeh and Fadi Dornaika and Abdelmalik Moujahid},
5+
doi = {10.1145/3718093},
6+
issn = {2157-6904},
7+
journal = {ACM Trans. Intell. Syst. Technol.},
8+
keywords = {Breast cancer classification, Deep learning, Angular margin-based softmax losses, Histopathological image, Discriminative feature embedding},
9+
publisher = {Association for Computing Machinery},
10+
title = {Cosine Additive Angular Margin Loss for Breast Cancer Classification in Histopathological Images with Small AI Systems},
11+
url = {https://doi.org/10.1145/3718093},
12+
year = {2025}
13+
}

‎content/publication/alirezazadeh-2025-cosine/index.md‎

Lines changed: 37 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,37 @@
1+
---
2+
title: Cosine Additive Angular Margin Loss for Breast Cancer Classification in Histopathological
3+
Images with Small AI Systems
4+
authors:
5+
- Pendar Alirezazadeh
6+
- Fadi Dornaika
7+
- Abdelmalik Moujahid
8+
date: '2025-01-01'
9+
publishDate: '2026-09-09T07:58:31.742348Z'
10+
publication_types:
11+
- article-journal
12+
publication: '*ACM Trans. Intell. Syst. Technol.*'
13+
doi: 10.1145/3718093
14+
abstract: Breast cancer claims thousands of lives annually, emphasizing the need for
15+
swift and accurate histopathology image classification to expedite diagnoses. Despite
16+
the rapid evolution of Convolutional Neural Networks (CNNs), the conventional softmax
17+
loss lacks the robustness required to discern intricate features in breast cancer
18+
classification from histopathological images. In response, our study introduces
19+
the Cosine Additive Angular Margin Loss (CAAM) to address this limitation and achieve
20+
both enhanced intra-class cohesion and distinct inter-class boundaries concurrently.
21+
By normalizing weight and feature vectors, we eliminate radial variations before
22+
imposing angular and cosine margin constraints on the softmax angle space. This
23+
process maximizes the decision margin, resulting in a discriminative feature embedding.
24+
Extensive experiments conducted on the BreakHis dataset demonstrate that CAAM consistently
25+
outperforms existing methods in breast cancer histopathological image classification.
26+
Our findings underscore the efficacy of angular margin-based softmax losses in bolstering
27+
the performance of advanced CNN models for breast cancer classification.
28+
tags:
29+
- Breast cancer classification
30+
- Deep learning
31+
- Angular margin-based softmax losses
32+
- Histopathological image
33+
- Discriminative feature embedding
34+
links:
35+
- name: URL
36+
url: https://doi.org/10.1145/3718093
37+
---

0 commit comments

Comments
 (0)