University of Salento

3 related articles

Subscribe to updates

researchsquare.com Oct 19, 2026
An Arabic chatbot folds depression screening into conversation, but the wrong embedder drops retrieval MAP to 4%
Rafiq is an Arabic conversational system for early depression detection that combines per-message symptom monitoring, a three-phase PHQ-9 questionnaire delivered inside the chat, and retrieval-augmented answers grounded in a clinical knowledge base. On 415 held-out instances symptom detection reached 0.98 accuracy, while retrieval quality depended heavily on embedder choice: an Arabic-specific model scored 80% MAP and a general multilingual mixture-of-experts model only 4%.
Rafiq is an Arabic conversational system for early depression detection that combines per-message symptom monitoring, a three-phase PHQ-9 questionnaire delivered inside the chat, and retrieval-augmented answers grounded in a clinical knowledge base. On 415 held-out instances symptom detection reached 0.98 accuracy, while retrieval quality depended heavily on embedder choice: an Arabic-specific model scored 80% MAP and a general multilingual mixture-of-experts model only 4%.
University of Salento Islamic University of Gaza
Alaa Alshurafa（First author）Motaz K H Saad（Corresponding author）
Rafiq Depression Detection PHQ-9 Screening Retrieval-Augmented Generation Arabic Mental Health Benchmark Dataset
researchsquare.com Oct 15, 2026
Adriatic reanalysis fields get a diffusion upscale: 1/12° input to 1/24° output, salinity error down to 0.35 psu
High-resolution Adriatic ocean reanalysis fields previously came from bilinear interpolation or costly nested regional models, and interpolation smooths away fronts and mesoscale structure. OcDiﬀSR applies a conditional denoising diffusion probabilistic model to jointly super-resolve sea-surface temperature, salinity, and horizontal velocity, using 1/12° GLORYS12V1 fields as conditioning input to reconstruct 1/24° Med MFC fields. On the independent test year 2009, scalar-field errors fall well below interpolation and the residual diffusion model CorrDiﬀ, with salinity RMSE of 0.346 psu and temperature RMSE of 0.477°C, while velocity correlation stays near 0.64.
High-resolution Adriatic ocean reanalysis fields previously came from bilinear interpolation or costly nested regional models, and interpolation smooths away fronts and mesoscale structure. OcDiﬀSR applies a conditional denoising diffusion probabilistic model to jointly super-resolve sea-surface temperature, salinity, and horizontal velocity, using 1/12° GLORYS12V1 fields as conditioning input to reconstruct 1/24° Med MFC fields. On the independent test year 2009, scalar-field errors fall well below interpolation and the residual diffusion model CorrDiﬀ, with salinity RMSE of 0.346 psu and temperature RMSE of 0.477°C, while velocity correlation stays near 0.64.
University of Salento
Rajat Srivastava（First author）Italo Epicoco（Corresponding author）
Oceanographic Super-Resolution Conditional Diffusion Model Statistical Downscaling
arXiv Oct 5, 2026
Quantum reinforcement learning for 5G and beyond energy saving: simulations show faster convergence and lower base-station energy use while maintaining QoS
Addressing the fact that base stations consume roughly 60-70% of radio access network energy in 5G and beyond networks, this article introduces a quantum reinforcement learning algorithm that uses superposition and entanglement through parameterized quantum circuits to optimize base-station energy use under dynamic user equipment behavior; simulations indicate it can substantially reduce energy consumption while maintaining quality of service and consistently outperforms deep reinforcement learning and Q-learning in convergence speed and learning complexity.
Addressing the fact that base stations consume roughly 60-70% of radio access network energy in 5G and beyond networks, this article introduces a quantum reinforcement learning algorithm that uses superposition and entanglement through parameterized quantum circuits to optimize base-station energy use under dynamic user equipment behavior; simulations indicate it can substantially reduce energy consumption while maintaining quality of service and consistently outperforms deep reinforcement learning and Q-learning in convergence speed and learning complexity.
Innovation Engineering (Italy)University of Salento
Muhammad Usman（First author）
Quantum Reinforcement Learning Energy Saving in 5G Networks

You’re all caught up

An Arabic chatbot folds depression screening into conversation, but the wrong embedder drops retrieval MAP to 4%

Adriatic reanalysis fields get a diffusion upscale: 1/12° input to 1/24° output, salinity error down to 0.35 psu

Quantum reinforcement learning for 5G and beyond energy saving: simulations show faster convergence and lower base-station energy use while maintaining QoS