62880 research outputs found
Sort by
Maize monoculture supported pre-Columbian urbanism in southwestern Amazonia
Get PDFThe Casarabe culture (500–1400 ce), spreading over roughly 4,500 km2 of the monumental mounds region of the Llanos de Moxos, Bolivia, is one of the clearest examples of urbanism in pre-Columbian (pre-1492 ce) Amazonia. It exhibits a four-tier hierarchical settlement pattern, with hundreds of monumental mounds interconnected by canals and causeways1,2. Despite archaeological evidence indicating that maize was cultivated by this society3, it is unknown whether it was the staple crop and which type of agricultural farming system was used to support this urban-scale society. Here, we address this issue by integration of remote sensing, field survey and microbotanical analyses, which shows that the Casarabe culture invested heavily in landscape engineering, constructing a complex system of drainage canals (to drain excess water during the rainy season) and newly documented savannah farm ponds (to retain water in the dry season). Phytolith analyses of 178 samples from 18 soil profiles in drained fields, farm ponds and forested settings record the singular and ubiquitous presence of maize (Zea mays) in pre-Columbian fields and farm ponds, and an absence of evidence for agricultural practices in the forest. Collectively, our findings show how the Casarabe culture managed the savannah landscape for intensive year-round maize monoculture that probably sustained its relatively large population. Our results have implications for how we conceive agricultural systems in Amazonia, and show an example of a Neolithic-like, grain-based agrarian economy in the Amazon
Adaptive monitoring for multimode nonstationary processes using cointegration analysis and probabilistic slow feature analysis
Get PDFThe condition monitoring of nonlinear, nonstationary and multimode processes is a difficult problem. Traditional multimode process monitoring methods generally assume that data from all potential modes are available, yet new modes may appear continuously in practice. This paper investigates an intelligent adaptive monitoring method for multimode nonstationary processes, which can deal with
the appearance of new modes with ease. A full-condition comprehensive framework is proposed to decompose feature subspaces. First, long-term equilibrium features are extracted by adaptive integration analysis (ACA) to identify the mode, without using any prior mode information intelligently for online applications. Then, recursive attention probabilistic slow feature analysis integrated with elastic weight consolidation (RAttPSFA-EWC) is investigated to deal with the remaining dynamic information and extract dynamic and static slow features to maintain continual learning for multimodes. Once a new mode is detected automatically, the previously learned knowledge is consolidated while extracting new features, which is beneficial to enhancing the performance of similar modes. The proposed ACA-RAttPSFA-EWC acts as online adaptive method by parameter updates with incoming normal data. Furthermore, several advanced methods are compared to demonstrate the strengths of ACA-RAttPSFA-EWC, and the proposed method is validated to be effective using a numerical case and a practical system
Eyes on the prize: tracking electron transfer in G‑rich duplex and quadruplex DNA using enantiopure ruthenium polypyridyl infrared redox probes
Get PDFPhotosensitized damage by the mechanism of direct 1e− transfer from a nucleobase to the metal complex is a
complementary approach to type I and type II methods of
photodynamic therapy. In this ultrafast spectroscopic study we report the ability of a nitrile infrared redox probe to report on the photo-oxidation of guanine-rich DNA, comprising persistent runs of guanine, by the dppz-10-CN containing complex [Ru(TAP)2(dppz-10-CN)]2+ (12+), dppz-10-CN = 10-cyano-dipyrido[3,2-a:2′,3′-c]phenazine and TAP = 1,4,5,8-tetraazaphenanthrene. Our study reveals the ability of the enantiomers of 12+ to photo-oxidize guanine in double-stranded and quadruplex DNA. Transient visible
absorption reveals a high yield of the formation of the photoreduced metal complex due to photo-oxidation of guanine in the quadruplexbound 12+ systems, and that this is greater for the Λ enantiomer. Spectro-electrochemical and computational studies indicate the role of the dppz-10-CN as the preferred site of reduction, while time-resolved electronic absorption (TrA) spectroscopy highlights the
impact of the enantiomers on the yield of photo-oxidation in the DNA systems. Notably, time-resolved infrared (TRIR)
spectroscopy allows comprehensive tracking of the photo-oxidation dynamics by monitoring four key components, namely: (1) the transient band of the Ru/TAP-based lowest 3MLCT excited state, (2) bleach bands associated with DNA bases in close proximity to the excited state “site effect”, (3) the guanine radical cation band at ca. 1700 cm−1 and (4) the amplification of the red-shifted nitrile stretching vibration of the transient dppz-reduced complex. Together, these results allow detailed profiling of photoinduced electron transfer in DNA-bound ruthenium(II) polypyridyl complex systems and highlight the potential of such redox probes. Overall, this study presents an important insight regarding the nature of charge transfer in a Hoogsteen-bound guanine quadruplex compared to Watson-Crick GC base pairings
Listening in a noisy world: the impact of acoustic cues and background music on speech perception in autism
Get PDFRecognising speech in noise involves focusing on a target speaker while filtering out competing voices and sounds. Acoustic cues, like vocal characteristics and spatial location, help differentiate between speakers. However, autistic individuals may process these cues differently, making it more challenging for them to perceive speech in such conditions. This study investigated how autistic individuals use acoustic cues to follow a target speaker and whether background music increases processing demands. Thirty-six autistic and 36 non-autistic participants identified information from a target speaker while ignoring a competing speaker and background music. The competing speaker’s gender and location either matched or differed from the target. The autistic group exhibited lower mean accuracy across cue conditions, indicating general challenges in recognising speech in noise. Trial-level analyses revealed that while both groups showed accuracy improvements over time without acoustic cues, the autistic group demonstrated smaller gains, suggesting greater difficulty in tracking the target speaker without distinct acoustic features. Background music did not disproportionately affect autistic participants but had a greater impact on those with stronger local processing tendencies. Using a naturalistic paradigm mimicking real-life scenarios, this study provides insights into speech-in-noise processing in autism, informing strategies to support speech perception in complex environments
Methods to analyse nutrition in high dimensions
Get PDFNutrition is fundamental to all living systems. In animals, morphological, physiological, and behavioral
adaptations for acquiring and digesting food shape their responses to the environment. While experimental frameworks have advanced our understanding of animal nutrition, methodological tools to address
its complexity remain limited. This thesis introduces pioneering analytical methods to explore nutritional trade-offs and optimal diet balances using Geometric Framework for Nutrition (GF) performance
landscapes. Chapter 1 traces the historical development of GF, highlighting its independent origins in
agricultural sciences and evolutionary ecology. Chapter 2 presents the Vector of Positions approach—the
only method capable of analysing n-dimensional performance landscapes—validated using two landmark
datasets. Chapter 3, the first study in the Nutrigonometry series, introduces a computationally efficient algorithm to estimate nutritional trade-offs in three-dimensional performance landscapes—the most
common in the field. Using the Pythagorean theorem, I demonstrate its robustness across multiple
datasets, showing that ordinary linear regression outperforms machine learning models for these calculations. Chapter 4 provides the first systematic test of the GF experimental design, proposing improved
methods for high-dimensional nutritional experiments to enhance accuracy in estimating optimal diets
and performance landscapes. Chapter 5 applies differential geometry to measure curvature in performance
landscapes and introduces the Hausdorff distance for comparing performance landscapes. Chapter 6 integrates previous analytical methods to examine the evolution of optimal diets between sexes in insects.
My findings reveal that the protein-to-carbohydrate ratio is more similar among closely related species
and that sexual conflict over nutrition is evolutionarily conserved. This study represents the first empirical work in precision comparative nutrition. Chapter 7 applies Thales’ theorem of inscribed triangles
to analyse how animals cope with imbalanced diets. By triangulating food intake under suboptimal
conditions, I quantify deviations from Thales’ theorem, revealing underlying nutritional strategies. This
thesis advances our understanding of nutritional constraints in optimizing fitness. Through a range of
methodological advancements, my work provides new insights into how animal nutrition can be studied
in its full multidimensional complexity. These methods and applications have broad implications, and I
hope they will be extended to other areas of biology to further our understanding of how living organisms
interact with their diet and how nutritional responses evolve across the tree of life
Essays on deep learning in asset pricing
No full textThis thesis studies applications of deep learning in asset pricing. The thesis contributes to
the prediction of equity returns, portfolio weights, and realised volatility, as well as to the
interpretability of models through the measurement of variable importance, and the optimisation of hyperparameters using deep learning. This thesis presents three contributions.
The first contribution is that we analyse forecast combination methods in the context of machine learning to predict equity returns. Whilst individual models lack robustness, forecast
combinations over two levels display stability and have Sharpe ratios of up to 3.06 on data
from 1987 to 2020. We use decision trees in genetic algorithms to analyse the structure
of variable influence. The impact of these variables displays non-constancies and shows
variations across different models and data. We propose a new performance measure for
risk premium forecasts which leads to more robust evaluations than existing performance
measures such as R2, whilst providing economic interpretability. This measure can be linked
to the advantages models offer for portfolio choice.
The second contribution we make is that we propose a new way of highlighting the context
of information from past and present financial data to price equities in a universal approximation setup. The combination of adding a time wise lookback space and using contextual
deep learning architectures with convolutional-transformer architecture enhances pricing
in several ways. Using US stocks between 2003 and 2020, we improve the accuracy and
performance of the models, increasing Sharpe ratios of managed portfolios from 0.91 in
a simple feed forward benchmark to over 1.19 in our approach. The methodology gives
insights into the factors that underpin pricing in the time and predictor space, making mid-term momentum factors less and short term momentum factors as well as financial statement
scores more important than in the traditional approach.
The third contribution we make is that we propose a deep learning framework for asset
pricing that leverages multidimensional, mixed-frequency data with different durations,
accounting for time-lags in non-lead frequencies. Our analysis forecasts weekly returns,
realised volatilities, and optimal portfolio weights for the 500 largest US stocks using 143
financial indicators at weekly, monthly, and quarterly intervals. While our multi-kernel
architecture offers flexibility in handling irregularly timed data, it did not surpass simpler
models that use a unified frequency, which achieved a strong Sharpe ratio of 1.36. Despite
efforts in hyperparameter optimization, the model’s complexity limited its performance,
highlighting the potential for further enhancement in modelling mixed-frequency data
Evaluating the scoring system of an AI-integrated app to assess foreign language phonological decoding
Get PDFPhonological decoding in a foreign language (FL)—a two-part process involving first the ability to map written symbols to their corresponding sounds and second to pronounce them intelligibly—is foundational for reading and vocabulary acquisition. Yet assessing this skill efficiently and at scale in young learners remains a persistent challenge. Here, we introduce and evaluate the accuracy and effectiveness of a novel method for assessing FL phonological decoding using an AI-driven app that automatically scores children's pronunciation of symbol-sound correspondences. In a study involving 254 learners of French and Spanish (aged 10–11) across five UK primary schools, pupils completed a read-aloud task (14 symbol-sound correspondences) that was scored by the app’s automatic speech recognition (ASR) technology. The validity of these automated scores was tested by fitting them as independent variables in regression models predicting human auditory coding. The multiple significant relationships between automated and human scores that were established indicate that there is great potential for ASR-based tools to reliably assess phonological decoding in this population. These findings provide the first large-scale empirical validation of an AI-based assessment of FL decoding in children, opening new possibilities, applicable to a range of languages being learnt, for scalable and efficient assessment
Can a major geomagnetic and auroral disturbance originate from a solar active region close to the limb?
Get PDFKey Points
Eruption of a coronal mass ejection (CME) from an active region on the solar limb may be able cause major terrestrial space weather disturbance
Comparison of recent events observed using modern equipment gives key insights on great historic events such as the role of substorms
The longitudes at which low‐latitude aurora is seen depends on the Universal Time of the CME impact on Earth's magnetospher
