446494 research outputs found
Sort by
Occupancy-Aware Spatio-Temporal Building Energy Forecasting with a Hybrid Long ShortTerm Memory and Graph Neural Network Benchmark Using Public Datasets
No full textAccurate short-term forecasting of building energy demand is complicated by coupled temporal dynamics, cross-meter spatial effects, and occupancy-driven variability. We present an occupancy-aware spatiotemporal framework that uses a Long Short-Term Memory (LSTM) branch and a Graph Neural Network (GNN) branch, augmented with calibrated occupancy probabilities transferred from labeled sources to public corpora lacking occupancy labels. Using BDG2 and ASHRAE GEPIII, we construct physical, correlation kNN, and learned kNN graphs; engineer calendar– weather–lag/rolling features; and evaluate with forward-chaining splits across horizons t+1…t+24. Primary (MAE, RMSE, MAPE) and domain metrics (CVRMSE, NMBE) follow ASHRAE Guideline 14. The hybrid attains RMSE 2.766 kWh (BDG2) and 2.740 kWh (ASHRAE GEPIII), yielding 33.44% and 33.52% reductions versus a ridge/XGBoost baseline, and statistical parity with LSTM-only (ΔRMSE −0.23% on BDG2; +0.02% on ASHRAE GEPIII; paired tests p<0.05). Horizon-wise curves show stable gains—especially during business hours—and learned kNN typically provides the lowest average error. Per-meter distributions indicate 100% of meters satisfy CVRMSE ≤ 30% and ∣NMBE∣ ≤ 10%, supporting calibration and retro-commissioning use. These findings demonstrate that using temporal and graph-based spatial cues with transferable occupancy signals delivers robust, label-efficient multi-meter forecasting, with units standardized (kWh, °C) and |NMBE| consistently denoted for clarity
Modelling Bitcoin Price Volatility and The Bitcoin Mining Dilemma on Global Health
No full textIn the digital age, Bitcoin remains the first and most notable cryptocurrency. Over the years, its value has increased, making it a desirable digital asset with millions of enthusiasts who trade and invest daily. Bitcoin is highly volatile in comparison with traditional assets and in absolute terms. Understanding its volatility history helps investors decide whether to buy, sell, or hold. A mathematical model that accounts for volatility is essential for these decisions. Unfortunately, Bitcoin’s vast profit potential for investors comes with the dilemma of its negative impact on global environmental health, which needs serious attention. This study aims to model Bitcoin’s return volatility that can support investment decisions and, on the other hand, the negative impact of Bitcoin mining and outline the actions necessary to mitigate it
Artificial Intelligence-Based Intelligent Energy Management for Sustainable Reduction in Electricity Usage
No full textThese pressures have been heightened by the increasing volatility in power markets and growing demand for electricity globally. This work presents an AI-powered energy management framework that consists of three major components: (i) LSTM networks for accurate demand forecasting, (ii) Random Forest classifiers for robust anomaly detection, and (iii) a reinforcement learning-based scheduling algorithm for dynamic load optimization. Unlike existing works, heavily relying on IoT-integrated infrastructures, the proposed system performs effectively with legacy metering data, thus enhancing scalability while reducing deployment costs. Experiments on real-world consumption datasets demonstrate key performance gains: peak load reduction by 18%, savings in operational costs by 14%, overall energy efficiency improvement by 21%, and 96% anomaly detection accuracy. The obtained results confirm the validity of integrating forecasting, anomaly detection, and intelligent scheduling in one unified data-centric framework. The proposed solution offers an efficient, adaptive approach that is environmentally friendly for optimizing electricity usage in residential, commercial, and industrial settings
Contribution of hydrometric measurements to the analysis of hydrological functioning in an ungauged watershed in the context of climate variability: The case of the Oued M'Hasser springs (2022-2023)
No full textLa situation actuelle des ressources hydriques au Maroc notamment dans les régions montagneuses semi-arides constitue un enjeu et un défi majeur, particulièrement dans un contexte caractérisé par variabilité climatique prononcée. Combinée à la surexploitation, cette variabilité rend les ressources hydriques de plus en plus fragiles et vulnérables. Pour cette raison, les données hydrométriques sont devenues essentielles pour une gestion rationnelle des ressources en eau à l'échelle du bassin versants. Cependant, l'accès à ces données peut constituer un obstacle dans le cas des bassins versants non équipés de stations de mesure hydrométrique, à l'instar du bassin de l'Oued M'Hasser. C'est pourquoi cette étude vise à mener des campagnes de jaugeage de terrain visant à produire une base de données hydrométrique fiable, dans l'objectif est de construire une courbe de tarage permettant d'estimer les débits à différentes conditions hydrologiques. Ces mesures ont permis de caractériser les dynamiques hydrologiques saisonnières et d'identifier les réponses différentielles des sources aux fluctuations pluviométriques
Detection of AI-Generated Facial Images Using Convolutional Neural Networks
No full textIn the modern world, the artificial intelligence technology available enables the generation of human faces of which real world counterparts do not exist, and such potential offers a myriad of possibilities. Creativity can illustrate and fabricate work. Granted, technology of this nature can be wielded to serve the purpose of identity fraud, providing misinformation and other seemingly ‘immoral’ acts. Hence, this study aims to investigate the use of Convolutional Neural Networks (CNN) in composite face images created with ‘This person does not exist’ and ‘real life images’ download. Considering the study’s focus, the learning rate of 0.0001, sigmoid, 0.4 Dropout, and average pooling for tuning showed the desired learnt outcomes. The results were astounding, the model achieved 99% accuracy on validation and 97% accuracy on the training dataset. This accomplishment was attributed to a face’s underlying subtle features, such as its textures, symmetry, and visual interferences. Optimisation was conducted to measure generalisation, needing the model to perform on a new dataset with additional smartphone images. The accuracy was 84% for augmented and real images, with 5 outcomes correct of 6 sample images
Assessment of fish stock supplementation for sustainable production from reservoirs of Tamil Nadu state, India
No full textIndian reservoirs offer tremendous scope but fish production remains below potential. The present study evaluates the effectiveness of fish seed stocking on the yield of 62 reservoirs of Tamil Nadu, India during 2011–2020. The results indicate that all large (>50 km2), 64.1% of the medium (10−50 km2) and 50.7% of small (<10 km2) reservoirs are under-stocked with less than <50% of the recommended stocking density. The stocking efficiency was found to be lowest in large (0.05), medium (0.11) and highest in small reservoir (0.26). The regression analysis showed significant positive associations between stocking density and yield in small (R2=0.05), medium (R2=0.31) and negative in large (R2=0.008) reservoirs. The mean annual yield during the studied period was highest in small (137 kg ha−1 yr−1), followed by medium (86.64 kg ha−1 yr−1) and, large (46.1 kg ha−1 yr−1) reservoirs. The study indicated that insufficient availability of fingerlings and inconsistency in the seed quality as the major stumbling blocks in achieving the estimated production potential. It is suggested to improve the hatchery facility, adoption of enclosure culture for rearing fish fingerlings, creating awareness among the stakeholders to optimize the fish production from these resources and also for ensuring improved livelihood of the fishers dependent on this sector. This is the first long-term, state level evaluation of reservoir stocking in Tamil Nadu, complementing past national studies. These findings highlight the importance of optimal stocking density, advanced fingerling use, and cooperative-based management for sustaining fish production and fisher livelihoods in Tamil Nadu
On bisimulation in absence of restriction
No full textWe revisit the standard bisimulation equalities in process models free of the restriction operator. As is well-known, in general the weak bisimilarity is coarser than the strong bisimilarity because it abstracts from internal actions. In absence of restriction, those internal actions become somewhat visible, so one might wonder if the weak bisimilarity is still ‘weak’. We show that in CCScore (i.e., Milner’s standard CCS without τ-prefix, summation and relabelling) the weak bisimilarity indeed remains weak, i.e., still strictly coarser than the strong bisimilarity, even without the restriction operator. Essentially, this is due to the existence of the replication operation, which can keep a process retaining its state (i.e., the capacity of interaction). By virtue of these observations, we examine a variant of the weak bisimilarity, called quasi-strong bisimilarity. This quasi-strong bisimilarity requires the matching of internal actions to be conducted in the strong manner, as for the strong bisimilarity, and the matching of visible actions to have no trailing internal actions. We exhibit that in CCScore without the restriction operator, the weak bisimilarity exactly collapses onto this quasi-strong bisimilarity, which is moreover shown to coincide with the branching bisimilarity. These results reveal that in absence of the restriction operation, some ingredient of the weak bisimilarity indeed turns into strong, particularly the matching of internal actions
Apollon : une infrastructure de pointe pour explorer la physique de l’extrême
No full textApollon est une infrastructure laser d’Ultra-Haute Intensité fournissant des impulsions multi-PW pour la recherche en interaction laser-matière et physique des plasmas. Ouverte aux utilisateurs depuis 2021, elle comporte deux zones : SFA pour les interactions à très haute intensité et l’accélération d’ions, et LFA pour l’accélération d’électrons. Apollon a déjà dépassé 1022 W/cm2 et poursuit sa montée en puissance vers de nouveaux régimes QED
Polarization echoes from past nuclear activity in the quasi-periodic eruption source GSN 069
No full textContext. X-ray quasi-periodic eruptions (QPEs) are repeating, high-amplitude, soft X-ray bursts observed from the nuclei of a dozen nearby low-mass galaxies. Their origin remains a major puzzle in the physics of accretion variability. Observational data indicate that X-ray and/or optical tidal disruption events (TDEs) may precede QPE detections. Although both kinds of outburst are driven by supermassive black holes, they are more frequently detected in faded active galactic nuclei (AGNs), when the TDE is not happening in a dormant galaxy. In the case of the QPE discovery source, GSN 069, observations and simulations have revealed evidence of past nuclear activity, although it remains debated whether this activity arose from a past AGN phase or from an enhanced TDE rate.
Aims. We investigated the origin of the past nuclear activity in GSN 069.
Methods. Past AGN activity imprints detectable polarization in optical light, due to the expected delay between direct and scattered light. On 6 September 2019, we targeted GSN 069 with VLT/FORS2 in both imaging polarimetry and spectropolarimetry modes so that its optical polarization could be investigated while the first detected QPE phase was still active.
Results. We measured a rising polarization, from ∼0% to ∼1.5%, as moving away from the nucleus of GSN 069. This rise is probed to be intrinsic to the central engine, confirming the already detected extended emission line region (EELR) by integral field unit data.
Conclusions. The increasing radial polarization demonstrates a switched-off nucleus. The polarization angle traces an axis aligned with elongated [OIII], [NII], and Hα gas distributions, revealing an EELR that may be consistent with relic polarization cones, therefore suggesting the presence of a torus-like structure in the past. Thus, optical polarization echoes geometrically favor a faded AGN as the origin of the EELR rather than a past elevated TDE rate, although the latter cannot be excluded
Comparison between Android Applications and Class-I Sound Level Meters in SPL measurement performance
No full textAccurate measurement of sound pressure level (SPL) is a cornerstone for understanding acoustic communication, animal behavior, and ecological dynamics. Acoustic signals can be broadly categorized into broadband and tonal sounds. Broadband noise—spanning a wide frequency range—is useful for assessing general noise levels, while tonal sounds, defined by distinct harmonic frequencies, provide critical insight into subtle phenomena such as animal communication, echolocation, and specific patterns of noise pollution. Smartphone-based SPL measurement has emerged as an accessible and cost-effective alternative to conventional sound level meters (SLMs). However, previous evaluations of SPL applications have primarily relied on iOS platforms and broadband noise measurements, leaving a significant gap in understanding how these tools perform for tonal sound measurements on Android devices—the more widely used operating system. To address these limitations, our study evaluated top three freely available Android SPL applications—Bolden, Kewlsoft, and KTW—against a Bru¨el & Kjær Class I SLM. The assessment employed both white noise and tonal stimuli across frequencies representative of natural soundscapes, facilitating a comprehensive performance comparison. Among these, the Bolden Sound Level Meter consistently delivered the most accurate readings for tonal measurements, whereas KTW performed well for broadband noise measurement, demonstrating the potential of Android-based tools in advancing environmental monitoring and wildlife conservation initiatives