224092 research outputs found
Sort by
Explainable AI-aided feature selection and model reduction for DRL-based V2X resource allocation
No full textArtificial intelligence (AI) is expected to significantly enhance radio resource management (RRM) in sixth-generation (6G) networks. However, the lack of explainability in complex deep learning (DL) models poses a challenge for practical implementation. This paper proposes a novel explainable AI (XAI)-based framework for feature selection and model complexity reduction in a model-agnostic manner. Applied to a multi-agent deep reinforcement learning (MADRL) setting, our approach addresses the joint sub-band assignment and power allocation problem in cellular vehicle-to-everything (V2X) communications. We propose a novel two-stage systematic explainability framework leveraging feature relevance-oriented XAI to simplify the DRL agents. While the former stage generates a state feature importance ranking of the trained models using Shapley additive explanations (SHAP)-based importance scores, the latter stage exploits these importance-based rankings to simplify the state space of the agents by removing the least important features from the model’s input. Simulation results demonstrate that the XAI-assisted methodology achieves ~97% of the original MADRL sum-rate performance while reducing optimal state features by ~28%, average training time by ~11%, and trainable weight parameters by ~46% in a network with eight vehicular pairs.</p
End-to-end learning of beam probing and RSSI-based multi-user hybrid precoding design
No full textThis paper presents an end-to-end (E2E) autoencoder learning framework that relies on unsupervised deep learning for the joint design of millimeter wave (mmWave) probing beams and hybrid precoding matrices in multi-user communication systems. Our model utilizes prior channel observations to achieve two main objectives: designing a compact set of probing beams and predicting off-grid radio frequency (RF) beamforming vectors. The E2E learning framework optimizes probing beams in an unsupervised manner, concentrating sensing power on promising spatial directions based on the environment. To this aim, we develop a neural network architecture respecting RF chain constraints and model received signal strength (RSS) using complex-valued convolutional layers. The autoencoder is trained to directly produce RF beamforming vectors for hybrid architectures based on projected RSS indicators (RSSIs). Once RF beamforming vectors for multi-users are predicted, baseband digital precoders are designed by accounting for multi-user interference. The autoencoder neural network is trained E2E in an unsupervised manner with a customized loss function aimed at maximizing RSS. In a system with 64 antennas, 4 RF chains, and 4 users, our approach requires only 8 probing beams to design RF beamforming vectors, compared to the conventional predefined codebooks with 64 or 128 beams.</p
Men’s partner-objectification vs. women’s perceived partner-objectification in heterosexual couples: outcomes for women’s self-objectification, sexual self-consciousness, and orgasm frequency
No full textBelow I discuss my MSc in Health Psychology dissertation which explored partner-objectification in the context of heterosexual couples and received the 2024 DHP MSc Research Award
Estimates of irrigation water volume by assimilation of satellite land surface temperature or soil moisture into a water-energy balance model in Morocco
No full textThe agricultural sector is the biggest and least efficient water user, accounting for around 80% of total water use in North Africa, which is already strongly impacted by climate change with prolonged drought periods, imposing limitations on irrigation water availability. The objective of this study was to estimate irrigation water use for the irrigation district of Doukkala in Morocco from 2017 to 2022 at daily resolution. The approach is based on the energy-water balance model FEST-EWB, which computes continuously in time on a pixel basis the main processes of the hydrological cycle and models evapotranspiration and soil moisture (SM) dynamics in the agricultural soil layer by solving the energy and water mass balance equations. Three different approaches were implemented to quantify actual irrigation volumes: (a) FAO-approach with the irrigation scheduling based on soil moisture and crop stress thresholds, (b) assimilation of satellite land surface temperature (LST) (downscaled Sentinel-3 data) and (c) assimilation of satellite soil moisture (SMAP-Sentinel-1 data). The model was first calibrated over non-irrigated areas, against LST from LANDSAT and Sentinel-3. The three irrigation approaches were then validated against soil moisture and evapotranspiration from reference models (MOD16 and WaPOR). The assimilation of LST gave the best estimates of total irrigation volumes compared to observed water allocation data (relative error = 1.5%). The FAO approach also performed well but slightly overestimated the observed data by 15%. On the other hand, coarse pixel resolution and low revisit time affected the performance of the satellite SM assimilation (relative error of −80%)
‘Follow the Science’: Popular Trust in Scientific Experts During the Coronavirus Pandemic
No full textThe coronavirus pandemic increased the role played by scientific advisers in counselling governments and citizens on issues around public health. This raises questions about how citizens evaluate scientists, and in particular the grounds on which they trust them. Previous studies have identified various factors associated with trust in scientists, although few have systematically explored a range of judgements and their relative effects. This study takes advantage of scientific advisers’ heightened public profile during the pandemic to explore how people’s trust in scientists is shaped by perceptions about their features and traits, along with evaluations of their behaviour and role within the decision-making process. The study also considers people’s trust in politicians, thereby identifying whether trust in scientists reflects similar or distinctive considerations to trust in partisan actors. Data are derived from specially-designed conjoint experiments and surveys of nationally representative samples in Britain and the US
Children’s spaces of belonging in schools: bringing theories and stakeholder perspectives into dialogue
No full textThis paper discusses the question: What is the explanatory power of bringing into dialogue theories of space and place with participatory research approaches that focus on joint perspectives of pupils, teachers and researchers in understanding the dynamics of children’s places of belonging in schools? It advances an argument that understanding children’s spaces of belonging in schools is relatively limited, particularly from a theoretically sophisticated stance or from children’s perspectives. The paper concludes that bringing together concepts of relational space as analytical tools with a participatory approach can create a third space that challenges binary positioning of ‘in/out’ with the potential to act as a safe haven for reflection and growth.</p
Simultaneous wireless power and data transfer system with full-duplex mode based on half-cycle OFDM
No full textA simultaneous wireless power and data transfer system (SWPDT) with full-duplex mode based on half-cycle orthogonal frequency division multiplexing (OFDM) technology is proposed to suppress interference from the power and ipsilateral carriers. The power carrier is regarded as the subcarrier carrying all the “1” data, which is used as the reference frequency, and the data subcarrier frequencies are determined according to the orthogonal properties of the power and data carriers to achieve full-duplex transmission. Demodulation integration cycle is reduced to half of the fundamental wave period, which doubles the data transmission rate. Depending on the proposed OFDM-based modulation and demodulation scheme, there is no need to use additional topologies to suppress interference from the power carrier and crosstalk from ipsilateral data source. In this article, simplified models of power and data transmission channels are developed, analyzed, and optimized to improve the data transmission gain. Finally, a 220 W experimental prototype was built, and tests show that the full-duplex communication rate could reach 170 kbps, which validates the correctness and effectiveness of the proposed SWPDT system
In-situ shear modulus reduction with strain in stiff fissured clays and weathered mudstones
No full textThe non-linear stress-strain behaviour of stiff clays and weak rocks at small and medium strains may be a critical consideration in the design of geotechnical structures. Empirical methods have been developed for estimating the maximum shear modulus and the normalised shear modulus reduction with strain of fine-grained soils. These are usually expressed as functions of the void ratio (or specific volume) and average effective (confining) stress, based on results from laboratory tests. However, the fidelity of these equations has not been widely evaluated in-situ. This paper describes the use of in-situ measurements from an instrumented embankment to calculate the operational in-situ shear modulus of the underlying stiff clays and weathered mudstones at medium and large strains. It is shown that the shear modulus at very small strain of the weathered clays increased linearly with depth, consistent with empirical equations. The gradient of the normalised, non-linear stiffnesses of the clays were comparable with those measured in laboratory tests of fine-grained soils, at a range of strains. However, the values for the reference strain, where the maximum shear modulus reduces by 50%, were lower than was predicted by the empirical equations.<br/
Negative political identities and costly political action
No full textElite and mass level politics in many Western democracies is increasingly characterised by the expression of negative feelings towards political out-groups. While the existence of these feelings is well-documented, there is little evidence on the consequences of activating political identities during election campaigns. We test whether fundraising emails containing negative or positive political identity cues lead party supporters to donate money via a large pre-registered digital field experiment conducted in collaboration with a British political party. We find that emails containing negative as opposed to positive identity cues lead to a higher number and frequency of donations. We also find that negative identity cues were only effective when paired with an issue identity rather than a traditional party identity cue, resulting in a 15% increase in the probability of donating over the untreated control. Our results provide novel experimental evidence on the behavioural effects of activating identities in real-world political campaigns
Data sources:
Access Repository Dashboard
Do you manage e-Prints Soton? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!