Automatic prediction of emotion promises to revolutionise human-computer
interaction. Recent trends involve fusion of multiple data modalities - audio,
visual, and physiological - to classify emotional state. However, in practice,
collection of physiological data `in the wild' is currently limited to
heartbeat time series of the kind generated by affordable wearable heart
monitors. Furthermore, real-world applications of emotion prediction often
require some measure of uncertainty over model output, in order to inform
downstream decision-making. We present here an end-to-end deep learning model
for classifying emotional valence from unimodal heartbeat time series. We
further propose a Bayesian framework for modelling uncertainty over these
valence predictions, and describe a probabilistic procedure for choosing to
accept or reject model output according to the intended application. We
benchmarked our framework against two established datasets and achieved peak
classification accuracy of 90%. These results lay the foundation for
applications of affective computing in real-world domains such as healthcare,
where a high premium is placed on non-invasive collection of data, and
To optimize plasmid containment, we have systematically investigated the factors that limit the killing efficiency of a suicide system based on the relF gene from Escherichia coli controlled by inducible lac promoters and placed on plasmids. In induction experiments with this suicide system, killing efficiency was unaffected by temperature and growth medium; there was no requirement for great promoter strength or high plasmid copy number. We could demonstrate that the factors limiting killing were the mutation rate of the suicide function and the reduced growth rate caused by a basal level of expression of the suicide gene during normal growth, which can give a selective growth advantage to cells with mutated suicide functions. The capacity of the plasmid-carried killing system to contain the plasmid was tested in transformation, transduction, and conjugational mobilization. The rate of plasmid transfer detected in these experiments seemed too high to provide adequate biological containment. As expected from the induction experiments, plasmids that escaped containment in these transfer experiments turned out to be mutated in the suicide function. With lac-induced suicide as a test, the efficiency of the system was improved by tightening the repression of the suicide gene, thereby preventing selection of cells mutated in the killing function. Reduction of the mutational inactivation rate of the suicide system by duplication of the suicide function augmented the efficiency of the suicide dramatically. These results permit the construction of extremely efficient biological containment systems.
The navigational skills of ants, bees and wasps represent one of the most baffling examples of the powers of minuscule brains. Insects store long-term memories of the visual scenes they experience, and they use compass cues to build a robust representation of directions. We know reasonably well how long-term memories are formed, in a brain area called the Mushroom Bodies (MB), as well as how heading representations are formed in another brain area called the Central Complex (CX). However, how such memories and heading representations interact to produce powerful navigational behaviours remains unclear. Here we combine behavioural experiments with computational modelling that is strictly based on connectomic data to provide a new perspective on how navigation might be orchestrated in these insects. Our results reveal a lateralised design, where signals about whether to turn left or right are segregated in the left and right hemispheres, respectively. Furthermore, we show that guidance is a two-stage process: the recognition of visual memories, presumably in the MBs, does not directly drive the motor command, but instead updates a desired heading, presumably in the CX, which in turn is used to control guidance using celestial compass information. Overall, this circuit enables ants to recognise views independently of their body orientation, and combines terrestrial and celestial cues in a way that produces exceptionally robust navigation.
Together, our data indicate that light exerts opposite regulation of LAZY4 expression in shoots and roots by mediating the protein levels of PIFs and HY5, respectively, to inhibit the negative gravitropism of shoots and promote positive gravitropism of roots in Arabidopsis.
In hypocotyls, light promotes degradation of PIFs to reduce LAZY4 expression, which inhibits the negative gravitropism of hypocotyls.