@@ -1,25 +1,27 @@
-@article{lekunberri2026deep,
-  title={Deep learning for accurate vision-based catch composition in tropical tuna purse seiners.},
-  author={Xabier Lekunberri and Ahmad Kamal and Izaro Goienetxea and Jon Ruiz and Iñaki Quincoces and Jaime Valls Miro and Ignacio Arganda-Carreras and Jose A Fernandes-Salvador},
-  journal={ICES Journal of Marine Science},
-  volume={83},
-  number={8},
-  pages={fsag151},
-  year={2026},
-  doi={10.1093/icesjms/fsag151},
-  url={https://doi.org/10.1093/icesjms/fsag151},
-  abstract={Purse seiners play a crucial role in tuna fishing, as approximately 69\% of the world's tropical tuna is caught using this gear. All tuna Regional Fisheries Management Organizations have established minimum standards for the use of electronic monitoring (EM) in fisheries in addition to traditional observers. These EM systems produce a massive amount of video data that human analysts must process. Integrating artificial intelligence (AI) into their workflow can decrease that workload and improve the accuracy of the reports. However, species identification still poses significant challenges for AI, as achieving balanced performance across all species requires appropriate training data. Here, we quantify the difficulty experts face to distinguish bigeye tuna (BET, Thunnus obesus) from yellowfin tuna (YFT, Thunnus albacares) using images captured by EM systems. We found inter-expert agreements of 42.9\% ± 35.6\% for BET and 57.1\% ± 35.6\% for YFT. We then present a multi-stage pipeline to estimate the species composition of the catches using a reliable ground-truth dataset based on identifications made by observers on board. Three segmentation approaches are compared: Mask R-CNN, a combination of DINOv2 with SAM2, and an integration of YOLOv9 with SAM2. We found that the latter performs the best, with a validation mean average precision of 0.66 ± 0.03 and a recall of 0.85 ± 0.03. Segmented individuals are tracked using ByteTrack. For classification, we evaluated a standard multiclass classification model and a hierarchical approach. Their performance depended on the fishing trip considered, suggesting that both approaches are promising, but additional testing is needed to determine which strategy generalizes better overall. All our models were cross-validated during training and tested on fishing operations with fully known catch composition. Combining YOLOv9-SAM2 with the hierarchical classification produced the best estimations, with 84.8\% of the individuals being segmented and classified with a mean absolute error of 4.5\%.}
+@article{abou2025agentic,
+  title={Agentic AI: a comprehensive survey of architectures, applications, and future directions},
+  author={Mohamad Abou Ali and Fadi Dornaika and Jinan Charafeddine},
+  journal={Artificial Intelligence Review},
+  volume={59},
+  number={1},
+  pages={11},
+  year={2025},
+  publisher={Springer},
+  url={https://doi.org/10.1007/s10462-025-11422-4},
+  doi={10.1007/s10462-025-11422-4},
+  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.}
 }
 
-@article{dangelo2026three,
-  title={Three-dimensional mapping of intact ovaries reveals the aging dynamics of the ovarian reserve.},
-  author={Arturo D'Angelo and Daniel Franco-Barranco and Marco Musy and Juan Manuel Duran and James Sharpe and Nicholas Stroustrup and Ignacio Arganda-Carreras and Elvan Böke},
-  journal={Nature Aging},
-  volume={6},
-  number={8},
-  pages={1580--1591},
-  year={2026},
-  doi={10.1038/s43587-026-01178-z},
-  url={https://doi.org/10.1038/s43587-026-01178-z},
-  abstract={Female fertility depends on a finite pool of oocytes that depletes during aging1,2, yet the spatiotemporal dynamics of this depletion remain poorly understood. Traditional methods obscure the three-dimensional architecture of the ovary, limiting quantitative insights. Here we combine light-sheet microscopy, artificial intelligence-driven segmentation and mathematical modeling to map over 85,000 oocytes in whole ovaries across the reproductive lifespan in mouse. We find that newly activated oocytes represent a fixed fraction of the total oocyte pool despite an age-related decline in oocyte numbers. Spatial analysis revealed that oocytes are enriched along the lateral ovarian axis, and local oocyte density positively correlates with activation. We also uncover a bimodal distribution of oocyte sizes, suggesting a bottleneck during oogenesis. Finally, a differential equation-based model captures the kinetics of oocyte activation and loss. Our findings establish a quantitative framework for understanding ovarian aging and suggest that an organ-scale regulatory mechanism coordinates the age-related decline in oocyte numbers.}
+@article{Alirezazadeh2025cosine,
+author = {Pendar Alirezazadeh and Fadi Dornaika and Abdelmalik Moujahid},
+title = {Cosine Additive Angular Margin Loss for Breast Cancer Classification in Histopathological Images with Small AI Systems},
+year = {2025},
+publisher = {Association for Computing Machinery},
+address = {New York, NY, USA},
+issn = {2157-6904},
+url = {https://doi.org/10.1145/3718093},
+doi = {10.1145/3718093},
+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.},
+journal = {ACM Trans. Intell. Syst. Technol.},
+keywords = {Breast cancer classification, Deep learning, Angular margin-based softmax losses, Histopathological image, Discriminative feature embedding}
 }