150259 research outputs found
Sort by
Bio‐Inspired Multimodal Hardware Front‐End Enabled by 2D Floating‐Gate Memory for UAV Perception
Full text linkABSTRACT Reliable environmental perception for small autonomous unmanned aerial vehicles (UAVs) remains challenging under rapid ego‐motion, visual blind regions, and aerodynamic disturbances. Inspired by birds’ efficient sensing‐to‐computing pathways, we design a multimodal joint‐modulation hardware system in which a 2D floating‐gate (FG) memory serves as the computing core, integrating visual, inertial, and wind‐field cues to enable fast and stable tracking and obstacle avoidance in dynamic environments. We develop a MoS 2 /h‐BN/graphene FG device that provides stable multilevel conductance states, an on/off ratio above 10 8 , sub‐10 µs switching, long retention, and high device uniformity. A 4 × 4 FG‐memory array robustly encodes temporal visual variations for real‐time target tracking, while a single FG device acts as an airflow neuron that rapidly detects UAV‐induced airflow in visual blind regions. An inertial‐information‐driven adaptive threshold modulation scheme further stabilizes both pathways under rapid ego‐motion, enabling bird‐like tracking and avoidance. Experiments show that visual processing latency is ∼7 ms, the average tracking center offset rate is 11.5%, background drift suppression exceeds 80%, and airflow disturbances trigger avoidance within 2 ms. These results demonstrate that the proposed system significantly improves signal‐processing speed and robustness, enhancing UAV applicability in unstructured environments
Fitting a lattice model with local and global transmission to spread of a plant disease.
Full text linkUnderstanding, predicting and managing the spread of plant pathogens is crucial given the economic, societal and climatic benefits of plants, including crops and trees. Mathematical models have long been used to investigate disease dynamics in plants. An important component of such models is to account for spatial structure, since plant hosts are immobile and a majority of disease spread will often be localised. Here we apply a lattice-based mathematical modelling approach, a pair approximation, to model disease spread. While this method has previously been used to develop epidemiological theory, it has not been used to predict spread in a specific pathosystem. We fit our lattice-based epidemiological model to experimental data relating to Bahia bark scaling of citrus, an economically-important disease in north-eastern Brazil, and compare its performance to a more commonly used dispersal-kernel modelling approach. We show that the lattice-based model fits the data well, predicting a significant degree of near-neighbour infections, with similar estimated values of epidemiologically-meaningful parameters to the dispersal model. We highlight the pros and cons of the lattice-based approach and discuss how it may be used to predict disease spread and optimise control of plant diseases
Friend Requests from the Force: Affective Mimicry, Intimate Imitations and a Softened Police Apparatus
Full text linkWhat happens when we are spoken to softly, with a sense of playfulness, by the symbolic apparatus of the carceral state? Invoking the concept of “affective mimicry,” this paper examines the digital means by which law enforcement agencies attempt to realise a sense of trust, intimacy, and emotional entanglement with the cyber-public. With a focus on the social media presence of the Australian Police, it will be argued that a strategic synthesis of memes, humorous language, and innocuous imagery with incarcerations, mugshots, and criminal descriptors abstracts the very materiality of law enforcement; its role, its potential misuse of power, and, by extension, state-sanctioned violence. Furthermore, the paper will suggest that the techno-affective cues embedded within these digital posts are vital in actively fostering intimate, off-screen solidarities between civilian users and the police force – solidarities which are oriented towards the visible, and always accessible, criminal other (with haptic contact evoked through virtual commentary and reactions). It will be posited that these interactions exist as potential avenues for exculpation, with digital posts momentarily capturing an affect and subsequently utilising it to bolster state governance. In this case, it happens to be the appropriation of a pre-existing, emotive lexicon, one commonly circulated in the context of community and friendship. The paper will also draw upon Louis Althusser’s analysis of the Ideological State Apparatus by conceptualising digital affect as possessing an interpellative function, much like classical forms of subjectification by the state, whereby the police are imagined as one of the primary social actors transforming individuals into subjects. However, Althusser’s point will be complicated through a recognition of social media’s emphasis on the general public as arbiters and decision-makers, an emphasis facilitated by the structural hyper- connectivity of digital communication, as well as the symbolic ideological terrain within which popular social media platforms have emerged. Indeed, the developmental system within which digital interactions between the state and its subjects occur ultimately obfuscates police subjectification and complicity for harm enacted. Considering all of this, the question remains: do we respond to friend requests sent by the force? If so, do we accept or reject
Future Directions in Aggression Research: The Contribution of Technology-Integrated Operationalizations
Full text linkAggression is a complex behavior that is difficult to capture using traditional methods, such as questionnaires and lab-based aggression tasks. These methods present challenges due to social desirability bias and limitations in translating findings into real-world situations. In this review, we discuss how emerging technologies, including virtual reality, video games, hyperscanning, biosignal recording, ecological momentary assessment and social media analysis, offer improved construct and ecological validity and can contribute to the refinement of integrative theoretical models of aggression. We comprehensively address advantages (e.g., immersion, realistic simulation, real-time and context-sensitive data collection and interpersonal dynamics) and limitations of each technology compared to traditional methods and highlight remaining gaps in aggression research. Additionally, we examine aggressive behavior related to the emergence of new technologies in digital spaces, focusing particularly on cyberbullying and the metaverse. We also review machine learning approaches for detecting cyber-aggression on social media platforms. We propose shifting from static, individual-level assessments to dynamic, context-sensitive frameworks that capture aggression in real time, in more ecological settings and digital environments. This shift in operationalization holds the potential to advance theoretical understanding, guide future research and inform clinical and forensic interventions
Reimagining Social Robots as Recommender Systems: Foundations, Framework, and Applications
No full textPersonalization in social robots refers to the ability of the robot to meet the needs and/or preferences of an individual user. Existing approaches typically rely on large language models (LLMs) to generate context-aware responses based on user metadata and historical interactions or on adaptive methods such as reinforcement learning (RL) to learn from users’ immediate reactions in real time. However, these approaches fall short of comprehensively capturing user preferences–including long-term, short-term, and fine-grained aspects–, and of using them to rank and select actions, proactively personalize interactions, and ensure ethically responsible adaptations. To address the limitations, we propose drawing on recommender systems (RSs), which specialize in modeling user preferences and providing personalized recommendations. To ensure the integration of RS techniques is well-grounded and seamless throughout the social robot pipeline, we (i) align the paradigms underlying social robots and RSs, (ii) identify key techniques that can enhance personalization in social robots, and (iii) design them as modular, plug-and-play components. This work not only establishes a framework for integrating RS techniques into social robots but also opens a pathway for deep collaboration between the RS and HRI communities, accelerating innovation in both fields
A new proof of the bunkbed conjecture in the p↑1 limit
Full text linkFor a finite simple graph G, the bunkbed graph G ± is defined to be the product graph G □ K 2 . We will label the two copies of a vertex v ∈ V ( G ) as v − and v + . The bunkbed conjecture, posed by Kasteleyn, states that for independent bond percolation on G ± , percolation from u − to v − is at least as likely as percolation from u − to v + , for any u , v ∈ V ( G ) . Despite the plausibility of this conjecture, so far the problem in full generality remains open. Recently, Hutchcroft, Nizić-Nikolac, and Kent gave a proof of the conjecture in the p ↑ 1 limit. Here we present a new proof of the bunkbed conjecture in this limit, working in the more general setting of allowing different probabilities on different edges of G ±
A human working memory advantage for social network information
Full text linkAs a social species, humans live in complexly bounded social groups. In order to navigate these networks, humans rely on a set of social–cognitive processes, including social working memory. Here, we designed a novel network memory task to study working memory for social versus non-social network information across 241 participants (18–65 years) in a tightly controlled, preregistered study. We show that humans demonstrate a working memory advantage for social, relative to non-social, network information. We also observed a self-relevant positivity bias, but an ‘other’ negativity bias. These findings are interpreted in the context of an evolutionary need to belong to one’s social group, to identify risks to one’s social safety and to appropriately track one’s social status within a complex network of social relationships
What is the “right” geographic market definition?
Full text linkThis paper examines the “right” geographic definition of relevant markets by analyzing how excise tax pass-through varies with local competition in the retail gasoline market of a large metropolitan city. Using a natural experiment from three unanticipated and exogenous fuel tax hikes and detailed station-level price data, we show that average pass-through is invariant to the number of nearby competitors across various geographic definitions. This contrasts with theoretical predictions and prior island-based evidence, suggesting that the entire metropolitan area functions as a single market. Our findings challenge standard isodistance- or isochrone-based market delineations used in academic research and competition policy
Effect of masker temporal pattern, spectrum, and presentation level on speech identification.
Full text linkThe effects of background spectral and temporal structure and overall level on the masking of speech were assessed for sentences presented in a speech-shaped noise (SSN), harmonic complex tone (HCT) (repetition period = 4.55 ms), and iterated ripple noise (IRN) (delay = 4.55 ms). The speech + noise was presented at 50 and 80 dB SPL using signal-to-noise ratios (SNRs) from 0 to -15 dB in 5-dB steps. The noises had similar spectral envelopes, but the HCT and IRN had spectral dips between peaks corresponding to the harmonic frequencies, and the HCT also had temporal dips. Especially for the SNRs of -5 and -10 dB, speech identification was best for the HCT masker and worst for the SSN, for both overall levels. For the SSN, the SNR required for 50% correct (SNR-50) was higher (worse) at 80 dB than at 50 dB, consistent with poorer frequency selectivity at the higher level. For the HCT, SNR-50 values were lower for the higher level, consistent with a better ability to "listen in the dips" at the higher level. The results indicate that the relative benefit of spectral and temporal dips varies with level
Structure Learning for Directed Trees
Full text linkKnowing the causal structure of a system is of fundamental interest in many areas of sci-
ence and can aid the design of prediction algorithms that work well under manipulations to
the system. The causal structure becomes identifiable from the observational distribution
under certain restrictions. To learn the structure from data, score-based methods evaluate
different graphs according to the quality of their fits. However, for large, continuous, and
nonlinear models, these rely on heuristic optimization approaches with no general guaran-
tees of recovering the true causal structure. In this paper, we consider structure learning
of directed trees. We propose a fast and scalable method based on Chu–Liu–Edmonds’
algorithm we call causal additive trees (CAT). For the case of Gaussian errors, we prove
consistency in an asymptotic regime with a vanishing identifiability gap. We also introduce
two methods for testing substructure hypotheses with asymptotic family-wise error rate
control that is valid post-selection and in unidentified settings. Furthermore, we study the
identifiability gap, which quantifies how much better the true causal model fits the obser-
vational distribution, and prove that it is lower bounded by local properties of the causal
model. Simulation studies demonstrate the favorable performance of CAT compared to
competing structure learning methods