Search

Multiphasic architectures are found ubiquitously in biomolecular condensates and are thought to have important implications for the organisation of multiple chemical reactions within the same compartment. Many of these multiphasic condensates contain RNA in addition to proteins. Here, we investigate the importance of different interactions in multiphasic condensates comprising two different proteins and RNA using computer simulations with a residue-resolution coarse-grained model of proteins and RNA. We find that in multilayered condensates containing RNA in both phases, protein-RNA interactions dominate, with aromatic residues and arginine forming the key stabilising interactions. The total aromatic and arginine content of the two proteins must be appreciably different for distinct phases to form, and we show that this difference increases as the system is driven towards greater multiphasicity. Using the trends observed in the different interaction energies of this system, we demonstrate that we can also construct multilayered condensates with RNA preferentially concentrated in one phase. The 'rules' identified can thus enable the design of synthetic multiphasic condensates to facilitate further study of their organisation and function.

Voting is used widely to identify a collective decision for a group of agents, based on their preferences. In this paper, we focus on evaluating and designing voting rules that support both the privacy of the voting agents and a notion of fairness over such agents. To do this, we introduce a novel notion of group fairness and adopt the existing notion of local differential privacy. We then evaluate the level of group fairness in several existing voting rules, as well as the trade-offs between fairness and privacy, showing that it is not possible to always obtain maximal economic efficiency with high fairness or high privacy levels. Then, we present both a machine learning and a constrained optimization approach to design new voting rules that are fair while maintaining a high level of economic efficiency. Finally, we empirically examine the effect of adding noise to create local differentially private voting rules and discuss the three-way trade-off between economic efficiency, fairness, and privacy. This paper appears in the special track on AI & Society.

Acknowledgements: We acknowledge funding from the University of Cambridge Ernest Oppenheimer Fund [PYC], the Winton Programme for the Physics of Sustainability [PYC, RC-G], the European Research Council under the European Union's Horizon 2020 research and innovation programme [grant 803326; RC-G]. JAJ is a Junior Research Fellow at King's College. This work was performed using resources provided by the Cambridge Tier-2 system operated by the University of Cambridge Research Computing Service funded by EPSRC Tier-2 capital grant EP/P020259/1 [RC-G, JAJ, AR]. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript., Intracellular condensates are highly multi-component systems in which complex phase behaviour can ensue, including the formation of architectures comprising multiple immiscible condensed phases. Relying solely on physical intuition to manipulate such condensates is difficult because of the complexity of their composition, and systematically learning the underlying rules experimentally would be extremely costly. We address this challenge by developing a computational approach to design pairs of protein sequences that result in well-separated multilayered condensates and elucidate the molecular origins of these compartments. Our method couples a genetic algorithm to a residue-resolution coarse-grained protein model. We demonstrate that we can design protein partners to form multiphase condensates containing naturally occurring proteins, such as the low-complexity domain of hnRNPA1 and its mutants, and show how homo- and heterotypic interactions must differ between proteins to result in multiphasicity. We also show that in some cases the specific pattern of amino-acid residues plays an important role. Our findings have wide-ranging implications for understanding and controlling the organisation, functions and material properties of biomolecular condensates., We acknowledge funding from the University of Cambridge Ernest Oppenheimer Fund [PYC], the Winton Programme for the Physics of Sustainability [PYC, RC-G], the European Research Council under the European Union’s Horizon 2020 research and innovation programme [grant 803326; RC-G]. JAJ is a Junior Research Fellow at King’s College. This work was performed using resources provided by the Cambridge Tier-2 system operated by the University of Cambridge Research Computing Service funded by EPSRC Tier-2 capital grant EP/P020259/1 [RC-G, JAJ, AR].

Language Models today provide a high accuracy across a large number of downstream tasks. However, they remain susceptible to adversarial attacks, particularly against those where the adversarial examples maintain considerable similarity to the original text. Given the multilingual nature of text, the effectiveness of adversarial examples across translations and how machine translations can improve the robustness of adversarial examples remain largely unexplored. In this paper, we present a comprehensive study on the robustness of current text adversarial attacks to round-trip translation. We demonstrate that 6 state-of-the-art text-based adversarial attacks do not maintain their efficacy after round-trip translation. Furthermore, we introduce an intervention-based solution to this problem, by integrating Machine Translation into the process of adversarial example generation and demonstrating increased robustness to round-trip translation. Our results indicate that finding adversarial examples robust to translation can help identify the insufficiency of language models that is common across languages, and motivate further research into multilingual adversarial attacks., Comment: Published at International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 2023

The era of the technological revolution brought rapid changes in the way businesses are managed and operated with new methods. This study applies and develops Artificial Neural Networks (ANNs) to predict the ROE (Return on Equity) and ROA (Return on Assets) based on data of the structure of the Board of Directors and managers of 839 Taiwanese electronics firms listed on the Taiwan Stock Exchange for the period 2000 to 2021. The results show that the characteristics of the Board of Directors and managers decide 64.25% of the value of the ROE and 67.05% of the ROA. Empirical results also show that the Board with fewer members is easy to reach a consensus on decisions rather than a larger board, leading to better firm performance. When ROE and ROA are at their worst, board members use their power to protect their wealth. However, independent board members have a negative influence on financial performance. Large company size has always been a strong supporter of high profitability, and a high debt ratio has not yet brought about tax savings.

Multiphasic architectures are found ubiquitously in biomolecular condensates and are thought to have important implications for the organisation of multiple chemical reactions within the same compartment. Many of these multiphasic condensates contain RNA in addition to proteins. Here, we investigate the importance of different interactions in multiphasic condensates comprising two different proteins and RNA using computer simulations with a residue-resolution coarse-grained model of proteins and RNA. We find that in multilayered condensates containing RNA in both phases, protein–RNA interactions dominate, with aromatic residues and arginine forming the key stabilising interactions. The total aromatic and arginine content of the two proteins must be appreciably different for distinct phases to form, and we show that this difference increases as the system is driven towards greater multiphasicity. Using the trends observed in the different interaction energies of this system, we demonstrate that we can also construct multilayered condensates with RNA preferentially concentrated in one phase. The ‘rules’ identified can thus enable the design of synthetic multiphasic condensates to facilitate further study of their organisation and function.

Understanding the thermodynamic stability and metastability of materials can help us to, for example, gauge whether crystalline polymorphs in pharmaceutical formulations are likely to be durable. It can also help us to design experimental routes to novel phases with potentially interesting properties. In this Perspective, we provide an overview of how thermodynamic phase behavior can be quantified both in computer simulations and machine-learning approaches to determine phase diagrams, as well as combinations of the two. We review the basic workflow of free-energy computations for condensed phases, including some practical implementation advice, ranging from the Frenkel-Ladd approach to thermodynamic integration and to direct-coexistence simulations. We illustrate the applications of such methods on a range of systems from materials chemistry to biological phase separation. Finally, we outline some challenges, questions, and practical applications of phase-diagram determination which we believe are likely to be possible to address in the near future using such state-of-the-art free-energy calculations, which may provide fundamental insight into separation processes using multicomponent solvents.

Adversarial examples causing evasive predictions are widely used to evaluate and improve the robustness of machine learning models. However, current studies focus on supervised learning tasks, relying on the ground-truth data label, a targeted objective, or supervision from a trained classifier. In this paper, we propose a framework of generating adversarial examples for unsupervised models and demonstrate novel applications to data augmentation. Our framework exploits a mutual information neural estimator as an information-theoretic similarity measure to generate adversarial examples without supervision. We propose a new MinMax algorithm with provable convergence guarantees for efficient generation of unsupervised adversarial examples. Our framework can also be extended to supervised adversarial examples. When using unsupervised adversarial examples as a simple plug-in data augmentation tool for model retraining, significant improvements are consistently observed across different unsupervised tasks and datasets, including data reconstruction, representation learning, and contrastive learning. Our results show novel methods and considerable advantages in studying and improving unsupervised machine learning via adversarial examples., This version has been accepted by AAAI-22

Establishing an analytical structure to classify proper criteria and identify the relative importance among them in different cities or municipalities for selection and resource allocations of government-sponsored regional innovation is crucial. In this paper, we construct an index system through the subject views of experts to evaluate regional scientific and technological innovation capabilities and collect empirical data to compare the six special municipalities in Taiwan by employing a hybrid CFPR-VIKOR approach. The research results show that a structural evaluation system is a basis for regional innovation policy. The need for fundamental data and regional resources assessments is apparent, and the role of the central government needs to be adapted to rapid changes. Through the design and adjustment of the regulatory system, decision-makers can decide how to allocate resources and further contribute to the development of regional innovation. We demonstrate that our approach can be extended to the regional innovation policies in the public sector. First published online 10 June 2022

Enterovirus A71 (EV-A71) causes major outbreaks of hand, foot, and mouth disease (HFMD) in many countries, most frequently affecting children, and a small proportion of cases may lead to death. Currently, no vaccine is available in most endemic regions, and no licenced treatments for EV-A71 infection are available. Here, we characterize a human monoclonal antibody (HuMAb), E1, by screening a Fab antibody phage library derived from patients who recovered from EV-A71 infection. E1 exhibits strong neutralizing activity against EV-A71 virus in cells. The cryo-electron microscopy (cryo-EM) structures of the EV-A71 virion in complex with E1 Fab fragments demonstrated that E1 recognized an epitope formed by residues in the BC and HI loops of VP1. In a mouse model, E1 effectively protected against lethal EV-A71 challenge in both prophylactic and therapeutic treatment. In particular, E1 significantly reduces virus titers and muscle damage. E1 might represent a potential adjunct to EV-A71 treatment.

RTL1/PEG11, which has been associated with anxiety disorders, is a retrotransposon-derived imprinted gene in the placenta. However, imprinting patterns and functions of RTL1 in the brain have not been well-investigated. We found Rtl1 was paternally, but not maternally, expressed in brain stem, thalamus, and hypothalamus of mice, and imprinting status of RTL1 was maintained in human brain. Paternal Rtl1 knockout (Rtl1m+/p-) mice had higher neonatal death rates due to impaired suckling, and low body weights beginning on embryonic day 16.5. High paternal expression of Rtl1 was detected in the locus coeruleus (LC) and Rtl1m+/p- mice showed an increased delay in time of onset for action potentials and inward currents with decreased neuronal excitability of LC neurons. Importantly, Rtl1m+/p- mice exhibited behaviors associated with anxiety, depression, fear-related learning and memory, social dominance, and low locomotor activity. Taken together, our findings demonstrate RTL1 is imprinted in brain, mediates emotional and social behaviors, and regulates excitability in LC neurons.

Probiotic supplements are potential therapeutic agents for age-related cognitive deficits.

Consider $K$ processes, each generating a sequence of identical and independent random variables. The probability measures of these processes have random parameters that must be estimated. Specifically, they share a parameter $\theta$ common to all probability measures. Additionally, each process $i\in\{1, \dots, K\}$ has a private parameter $\alpha_i$. The objective is to design an active sampling algorithm for sequentially estimating these parameters in order to form reliable estimates for all shared and private parameters with the fewest number of samples. This sampling algorithm has three key components: (i)~data-driven sampling decisions, which dynamically over time specifies which of the $K$ processes should be selected for sampling; (ii)~stopping time for the process, which specifies when the accumulated data is sufficient to form reliable estimates and terminate the sampling process; and (iii)~estimators for all shared and private parameters. Owing to the sequential estimation being known to be analytically intractable, this paper adopts \emph {conditional} estimation cost functions, leading to a sequential estimation approach that was recently shown to render tractable analysis. Asymptotically optimal decision rules (sampling, stopping, and estimation) are delineated, and numerical experiments are provided to compare the efficacy and quality of the proposed procedure with those of the relevant approaches.

Spreading depolarization (SD), the underlying mechanism of migraine aura, may trigger the opening of the pannexin 1 (PANX1) pore to sustain the cortical neuroinflammatory cascades involved in the genesis of headache. Yet, the mechanism underlying SD-evoked neuroinflammation and trigeminovascular activation remains incompletely understood. We characterized the identity of inflammasome activated following SD-evoked PANX1 opening. Pharmacological inhibitors targeting PANX1 or NLRP3 as well as genetic ablation of Nlrp3 and Il1b were applied to investigate the molecular mechanism of the downstream neuroinflammatory cascades. In addition, we examined whether SD-triggered microglial activation facilitates neuronal NLRP3-mediated inflammatory cascades. Pharmacological inhibition of toll-like receptors TLR2/4, the potential receptors of the damage-associated molecular pattern HMGB1, was further employed to interrogate the neuron-microglia interplay in SD-induced neuroinflammation. We found that NLRP3 but not NLRP1 or NLRP2 inflammasome was activated following PANX1 opening after single or multiple SDs evoked by either KCl topical application or non-invasively with optogenetics. The SD-evoked NLRP3 inflammasome activation was observed exclusively in neurons but not microglia or astrocytes. Proximity ligation assay demonstrated that the assembly of the NLRP3 inflammasome occurred as early as 15 min after SD. Genetic ablation of Nlrp3 or Il1b or pharmacological inhibition of PANX1 or NLRP3 ameliorated SD-induced neuronal inflammation, middle meningeal artery dilatation, calcitonin gene-related peptide expression in trigeminal ganglion and c-Fos expression in trigeminal nucleus caudalis. Moreover, multiple SDs induced microglial activation subsequent to neuronal NLRP3 inflammasome activation, which in turn orchestrated with neurons to mediate cortical neuroinflammation, as demonstrated by decreased neuronal inflammation after pharmacological inhibition of microglia activation or blockade of the TLR2/4 receptors. To conclude, single or multiple SDs evoked activation of neuronal NLRP3 inflammasomes and its downstream inflammatory cascades to mediate cortical neuroinflammation and trigeminovascular activation. In the context of multiple SDs, the cortical inflammatory processes could be facilitated by SD-evoked microglia activation. These findings may implicate the potential role of innate immunity in migraine pathogenesis.

Current studies on adversarial robustness mainly focus on aggregating local robustness results from a set of data samples to evaluate and rank different models. However, the local statistics may not well represent the true global robustness of the underlying unknown data distribution. To address this challenge, this paper makes the first attempt to present a new framework, called GREAT Score , for global robustness evaluation of adversarial perturbation using generative models. Formally, GREAT Score carries the physical meaning of a global statistic capturing a mean certified attack-proof perturbation level over all samples drawn from a generative model. For finite-sample evaluation, we also derive a probabilistic guarantee on the sample complexity and the difference between the sample mean and the true mean. GREAT Score has several advantages: (1) Robustness evaluations using GREAT Score are efficient and scalable to large models, by sparing the need of running adversarial attacks. In particular, we show high correlation and significantly reduced computation cost of GREAT Score when compared to the attack-based model ranking on RobustBench (Croce,et. al. 2021). (2) The use of generative models facilitates the approximation of the unknown data distribution. In our ablation study with different generative adversarial networks (GANs), we observe consistency between global robustness evaluation and the quality of GANs. (3) GREAT Score can be used for remote auditing of privacy-sensitive black-box models, as demonstrated by our robustness evaluation on several online facial recognition services.

Understanding the host-specificity of different families of viruses sheds light on the origin of, e.g., SARS-CoV-2, rabies, and other such zoonotic pathogens in humans. It enables epidemiologists, medical professionals, and policymakers to curb existing epidemics and prevent future ones promptly. In the family Coronaviridae (of which SARS-CoV-2 is a member), it is well-known that the spike protein is the point of contact between the virus and the host cell membrane. On the other hand, the two traditional mammalian orders, Carnivora (carnivores) and Chiroptera (bats) are recognized to be responsible for maintaining and spreading the Rabies Lyssavirus (RABV). We propose Virus2Vec, a feature-vector representation for viral (nucleotide or amino acid) sequences that enable vector-space-based machine learning models to identify viral hosts. Virus2Vec generates numerical feature vectors for unaligned sequences, allowing us to forego the computationally expensive sequence alignment step from the pipeline. Virus2Vec leverages the power of both the \emph{minimizer} and position weight matrix (PWM) to generate compact feature vectors. Using several classifiers, we empirically evaluate Virus2Vec on real-world spike sequences of Coronaviridae and rabies virus sequence data to predict the host (identifying the reservoirs of infection). Our results demonstrate that Virus2Vec outperforms the predictive accuracies of baseline and state-of-the-art methods., Comment: 11 Pages 6 Figures Accepted in conference Conference on Health, Inference, and Learning (CHIL) 2023

The bulk of existing research in defending against adversarial examples focuses on defending against a single (typically bounded Lp-norm) attack, but for a practical setting, machine learning (ML) models should be robust to a wide variety of attacks. In this paper, we present the first unified framework for considering multiple attacks against ML models. Our framework is able to model different levels of learner's knowledge about the test-time adversary, allowing us to model robustness against unforeseen attacks and robustness against unions of attacks. Using our framework, we present the first leaderboard, MultiRobustBench, for benchmarking multiattack evaluation which captures performance across attack types and attack strengths. We evaluate the performance of 16 defended models for robustness against a set of 9 different attack types, including Lp-based threat models, spatial transformations, and color changes, at 20 different attack strengths (180 attacks total). Additionally, we analyze the state of current defenses against multiple attacks. Our analysis shows that while existing defenses have made progress in terms of average robustness across the set of attacks used, robustness against the worst-case attack is still a big open problem as all existing models perform worse than random guessing., Comment: ICML 2023

Visual Prompting (VP) is an emerging and powerful technique that allows sample-efficient adaptation to downstream tasks by engineering a well-trained frozen source model. In this work, we explore the benefits of VP in constructing compelling neural network classifiers with differential privacy (DP). We explore and integrate VP into canonical DP training methods and demonstrate its simplicity and efficiency. In particular, we discover that VP in tandem with PATE, a state-of-the-art DP training method that leverages the knowledge transfer from an ensemble of teachers, achieves the state-of-the-art privacy-utility trade-off with minimum expenditure of privacy budget. Moreover, we conduct additional experiments on cross-domain image classification with a sufficient domain gap to further unveil the advantage of VP in DP. Lastly, we also conduct extensive ablation studies to validate the effectiveness and contribution of VP under DP consideration.

Diffusion Models (DMs) are state-of-the-art generative models that learn a reversible corruption process from iterative noise addition and denoising. They are the backbone of many generative AI applications, such as text-to-image conditional generation. However, recent studies have shown that basic unconditional DMs (e.g., DDPM and DDIM) are vulnerable to backdoor injection, a type of output manipulation attack triggered by a maliciously embedded pattern at model input. This paper presents a unified backdoor attack framework (VillanDiffusion) to expand the current scope of backdoor analysis for DMs. Our framework covers mainstream unconditional and conditional DMs (denoising-based and score-based) and various training-free samplers for holistic evaluations. Experiments show that our unified framework facilitates the backdoor analysis of different DM configurations and provides new insights into caption-based backdoor attacks on DMs.

Due to the significant computational challenge of training large-scale graph neural networks (GNNs), various sparse learning techniques have been exploited to reduce memory and storage costs. Examples include \textit{graph sparsification} that samples a subgraph to reduce the amount of data aggregation and \textit{model sparsification} that prunes the neural network to reduce the number of trainable weights. Despite the empirical successes in reducing the training cost while maintaining the test accuracy, the theoretical generalization analysis of sparse learning for GNNs remains elusive. To the best of our knowledge, this paper provides the first theoretical characterization of joint edge-model sparse learning from the perspective of sample complexity and convergence rate in achieving zero generalization error. It proves analytically that both sampling important nodes and pruning neurons with the lowest-magnitude can reduce the sample complexity and improve convergence without compromising the test accuracy. Although the analysis is centered on two-layer GNNs with structural constraints on data, the insights are applicable to more general setups and justified by both synthetic and practical citation datasets.

Recently, video summarization has been proposed as a method to help video exploration. However, traditional video summarization models only generate a fixed video summary which is usually independent of user-specific needs and hence limits the effectiveness of video exploration. Multi-modal video summarization is one of the approaches utilized to address this issue. Multi-modal video summarization has a video input and a text-based query input. Hence, effective modeling of the interaction between a video input and text-based query is essential to multi-modal video summarization. In this work, a new causality-based method named Causal Video Summarizer (CVS) is proposed to effectively capture the interactive information between the video and query to tackle the task of multi-modal video summarization. The proposed method consists of a probabilistic encoder and a probabilistic decoder. Based on the evaluation of the existing multi-modal video summarization dataset, experimental results show that the proposed approach is effective with the increase of +5.4% in accuracy and +4.92% increase of F 1- score, compared with the state-of-the-art method., Comment: This paper is accepted by IEEE International Conference on Multimedia and Expo (ICME), 2022

Deep neural networks (DNNs) have become ubiquitous in machine learning, but their energy consumption remains a notable issue. Lowering the supply voltage is an effective strategy for reducing energy consumption. However, aggressively scaling down the supply voltage can lead to accuracy degradation due to random bit flips in static random access memory (SRAM) where model parameters are stored. To address this challenge, we introduce NeuralFuse, a novel add-on module that addresses the accuracy-energy tradeoff in low-voltage regimes by learning input transformations to generate error-resistant data representations. NeuralFuse protects DNN accuracy in both nominal and low-voltage scenarios. Moreover, NeuralFuse is easy to implement and can be readily applied to DNNs with limited access, such as non-configurable hardware or remote access to cloud-based APIs. Experimental results demonstrate that, at a 1% bit error rate, NeuralFuse can reduce SRAM memory access energy by up to 24% while improving accuracy by up to 57%. To the best of our knowledge, this is the first model-agnostic approach (i.e., no model retraining) to address low-voltage-induced bit errors. The source code is available at https://github.com/IBM/NeuralFuse.

This paper addresses the task of anomaly diagnosis when the underlying data generation process has a complex spatio-temporal (ST) dependency. The key technical challenge is to extract actionable insights from the dependency tensor characterizing high-order interactions among temporal and spatial indices. We formalize the problem as supervised dependency discovery, where the ST dependency is learned as a side product of multivariate time-series classification. We show that temporal positional encoding used in existing ST transformer works has a serious limitation in capturing higher frequencies (short time scales). We propose a new positional encoding with a theoretical guarantee, based on discrete Fourier transform. We also propose a new ST dependency discovery framework, which can provide readily consumable diagnostic information in both spatial and temporal directions. Finally, we demonstrate the utility of the proposed model, DFStrans (Diagnostic Fourier-based Spatio-temporal Transformer), in a real industrial application of building elevator control.

Machine learning based methods have shown potential for optimizing existing molecules with more desirable properties, a critical step towards accelerating new chemical discovery. Here we propose QMO, a generic query-based molecule optimization framework that exploits latent embeddings from a molecule autoencoder. QMO improves the desired properties of an input molecule based on efficient queries, guided by a set of molecular property predictions and evaluation metrics. We show that QMO outperforms existing methods in the benchmark tasks of optimizing small organic molecules for drug-likeness and solubility under similarity constraints. We also demonstrate significant property improvement using QMO on two new and challenging tasks that are also important in real-world discovery problems: (i) optimizing existing potential SARS-CoV-2 Main Protease inhibitors toward higher binding affinity; and (ii) improving known antimicrobial peptides towards lower toxicity. Results from QMO show high consistency with external validations, suggesting effective means to facilitate material optimization problems with design constraints., Preprint version to be published at Nature Machine Intelligence; Github: https://github.com/IBM/QMO

Various physics- and data-driven sequence-dependent protein coarse-grained models have been developed to study biomolecular phase separation and elucidate the dominant physicochemical driving forces. Here we present Mpipi, a multiscale coarse-grained model that describes almost quantitatively the change in protein critical temperatures as a function of amino acid sequence. The model is parameterized from both atomistic simulations and bioinformatics data and accounts for the dominant role of π–π and hybrid cation–π/π–π interactions and the much stronger attractive contacts established by arginines than lysines. We provide a comprehensive set of benchmarks for Mpipi and seven other residue-level coarse-grained models against experimental radii of gyration and quantitative in vitro phase diagrams, demonstrating that Mpipi predictions agree well with experiments on both fronts. Moreover, Mpipi can account for protein–RNA interactions, correctly predicts the multiphase behavior of a charge-matched poly-arginine/poly-lysine/RNA system, and recapitulates experimental liquid–liquid phase separation trends for sequence mutations on FUS, DDX4 and LAF-1 proteins. Combining bioinformatics data and atomistic simulations, this study develops a sequence-dependent coarse-grained model for biomolecular phase separation. This model achieves a quantitative agreement with experimental observations. Extensive benchmarks exemplify its performance.