Search

Large Language Models (LLMs) play a crucial role in capturing structured semantics to enhance language understanding, improve interpretability, and reduce bias. Nevertheless, an ongoing controversy exists over the extent to which LLMs can grasp structured semantics. To assess this, we propose using Semantic Role Labeling (SRL) as a fundamental task to explore LLMs' ability to extract structured semantics. In our assessment, we employ the prompting approach, which leads to the creation of our few-shot SRL parser, called PromptSRL. PromptSRL enables LLMs to map natural languages to explicit semantic structures, which provides an interpretable window into the properties of LLMs. We find interesting potential: LLMs can indeed capture semantic structures, and scaling-up doesn't always mirror potential. Additionally, limitations of LLMs are observed in C-arguments, etc. Lastly, we are surprised to discover that significant overlap in the errors is made by both LLMs and untrained humans, accounting for almost 30% of all errors., Comment: Accepted by ICIC 2024

Deep learning models have shown promising performance for cell nucleus segmentation in the field of pathology image analysis. However, training a robust model from multiple domains remains a great challenge for cell nucleus segmentation. Additionally, the shortcomings of background noise, highly overlapping between cell nucleus, and blurred edges often lead to poor performance. To address these challenges, we propose a novel framework termed CausalCellSegmenter, which combines Causal Inference Module (CIM) with Diversified Aggregation Convolution (DAC) techniques. The DAC module is designed which incorporates diverse downsampling features through a simple, parameter-free attention module (SimAM), aiming to overcome the problems of false-positive identification and edge blurring. Furthermore, we introduce CIM to leverage sample weighting by directly removing the spurious correlations between features for every input sample and concentrating more on the correlation between features and labels. Extensive experiments on the MoNuSeg-2018 dataset achieves promising results, outperforming other state-of-the-art methods, where the mIoU and DSC scores growing by 3.6% and 2.65%., Comment: 10 pages, 5 figures, 2 tables, MICCAI

Since the strong comorbid similarity in NDDs, such as attention-deficit hyperactivity disorder, can interfere with the accurate diagnosis of autism spectrum disorder (ASD), identifying unknown classes is extremely crucial and challenging from NDDs. We design a novel open set recognition framework for ASD-aided diagnosis (OpenNDD), which trains a model by combining autoencoder and adversarial reciprocal points learning to distinguish in-distribution and out-of-distribution categories as well as identify ASD accurately. Considering the strong similarities between NDDs, we present a joint scaling method by Min-Max scaling combined with Standardization (MMS) to increase the differences between classes for better distinguishing unknown NDDs. We conduct the experiments in the hybrid datasets from Autism Brain Imaging Data Exchange I (ABIDE I) and THE ADHD-200 SAMPLE (ADHD-200) with 791 samples from four sites and the results demonstrate the superiority on various metrics. Our OpenNDD achieves promising performance, where the accuracy is 77.38%, AUROC is 75.53% and the open set classification rate is as high as 59.43%., Comment: This work has been submitted to the IEEE for possible publication

Short, partially complementary, single-stranded (ss)DNA strands can form nanostructures with a wide variety of shapes and mechanical properties. It is well known that semiflexible, linear dsDNA can undergo an isotropic to nematic (IN) phase transition and that sufficiently bent structures can form a biaxial nematic phase. Here we use numerical simulations to explore how the phase behaviour of linear DNA constructs changes as we tune the mechanical properties of the constituent DNA by changing the nucleotide sequence. The IN phase transition can be suppressed in so-called DNA 'nunchakus': structures consisting of two rigid dsDNA arms, separated by a sufficiently flexible spacer. In this paper, we use simulations to explore what phase behavior to expect for different linear DNA constructs. To this end, we first performed numerical simulations exploring the structural properties of a number of different DNA oligonucleotides, using the oxDNA package. We then used the structural information generated in the oxDNA simulations to construct more coarse-grained models of the rodlike, bent-core, and nunchaku DNA. These coarse-grained models were used to explore the phase behavior of suspensions of the various DNA constructs. The approach explored in this paper makes it possible to 'design' the phase behavior of DNA constructs by a suitable choice of the constituent nucleotide sequence.

'Soft' patchy surfactant micelles have become an additional building tool in self-assembling systems. The triblock copolymer, Pluronic F108, forms spherical micelles in aqueous solutions upon heating leading to a simple phase diagram with a micellar crystalline solid at higher temperatures and concentrations. Here we report the strong influence of end-functionalising the chain ends either with an azide or azide-DNA complex on the systems' phase behaviour. We find that the azide(N3)- functionalisation renders the chain ends weakly hydrophobic at lower temperatures, causing them to self-assemble into flower-micelles. This hydrophobicity increases with increasing temperature and poses a competing self-assembling mechanism to the solvent induces hydrophobic interactions between the middle-blocks of F108 at higher temperatures and leads to a macroscopic phase separation that is absent in the pure F108 system. However, when we attached short, hydrophilic single-stranded (ss)DNA to the azide groups via click chemistry the chain ends became 'sticky' due to DNA hybridisation below the melting temperature of the complementary ssDNA ends while reverting to hydrophilic behaviour above. We characterise their structural and rheological properties via Dynamic Light Scattering (DLS) and DLS-based passive microrheology with an improved time-frequency domain inversion step. We present the structural behaviour of dilute and semi-dilute solutions of the original F108 system and compare the results with solutions containing either the F108- azide (F108-N3) or partially DNA-functionalised F108-azide chains. Our DLS and microrheology studies inform us on how the attachment of azide groups on F108 changes the mechanical and structural properties of micellar fluids pioneering further characterisation and design of these hybrid systems., Comment: Ready to submit to Polymers - open access journal of polymer science published semimonthly online by MDPI. 23 pages, 8 figures

Three-dimensional DNA networks, composed of tri- or higher valent nanostars with sticky, single-stranded DNA overhangs, have been previously studied in the context of designing thermally responsive, viscoelastic hydrogels. In this work, we use linker-mediated gels, where the sticky ends of two trivalent nanostars are connected through the complementary sticky ends of a linear DNA duplex. We can design this connection to be either rigid or flexible by introducing flexible, non-binding bases. The additional flexiblity provided by these non-binding bases influences the effective elasticity of the percolating gel formed at low temperatures. Here we show that by choosing the right length of the linear duplex and non-binding flexible joints, we obtain a completely different phase behaviour to that observed for rigid linkers. In particular, we use dynamic light scattering as microrheological tool to monitor the self-assembly of DNA nanostars with linear linkers as a function of temperature. While we observe classical gelation when using rigid linkers, the presence of flexible joints leads to a cluster fluid with reduced viscosity. Using both the oxDNA model and a coarse-grained simulation to investigate the nanostar-linker topology, we hypothesise on the possible structure formed by the DNA clusters., Comment: 14 pages, 9 figures, 2 tables