Searcharxiv⌕ Search

arXiv subjects

Fan Bai

Publications and source records attributed to Fan Bai.

54 records · Page 3Linked to original sources
arXiv2021

Membraneless organelles formed by liquid-liquid phase separation increase bacterial fitness

Liquid-liquid phase separation is emerging as a crucial phenomenon in several fundamental cell processes. A range of eukaryotic systems exhibit liquid condensates. However, their function in bacteria, which in general lack membrane-bound compartments, remains less clear. Here, we used high-resolution optical microscopy to observe single bacterial aggresomes, nanostructured intracellular assemblies of proteins, to undercover their role in cell stress. We find that proteins inside aggresomes are mobile and undergo dynamic turnover, consistent with a liquid state. Our observations are in quantitative agreement with phase-separated liquid droplet formation driven by interacting proteins under thermal equilibrium that nucleate following diffusive collisions in the cytoplasm. We have discovered aggresomes in multiple species of bacteria, and show that these emergent, metastable liquid-structured protein assemblies increase bacterial fitness by enabling cells to tolerate environmental stresses.

q-bio.BM↗
arXiv2021

Federated Face Recognition

Face recognition has been extensively studied in computer vision and artificial intelligence communities in recent years. An important issue of face recognition is data privacy, which receives more and more public concerns. As a common privacy-preserving technique, Federated Learning is proposed to train a model cooperatively without sharing data between parties. However, as far as we know, it has not been successfully applied in face recognition. This paper proposes a framework named FedFace to innovate federated learning for face recognition. Specifically, FedFace relies on two major innovative algorithms, Partially Federated Momentum (PFM) and Federated Validation (FV). PFM locally applies an estimated equivalent global momentum to approximating the centralized momentum-SGD efficiently. FV repeatedly searches for better federated aggregating weightings via testing the aggregated models on some private validation datasets, which can improve the model's generalization ability. The ablation study and extensive experiments validate the effectiveness of the FedFace method and show that it is comparable to or even better than the centralized baseline in performance.

cs.CV↗
arXiv2021

Process-Level Representation of Scientific Protocols with Interactive Annotation

We develop Process Execution Graphs (PEG), a document-level representation of real-world wet lab biochemistry protocols, addressing challenges such as cross-sentence relations, long-range coreference, grounding, and implicit arguments. We manually annotate PEGs in a corpus of complex lab protocols with a novel interactive textual simulator that keeps track of entity traits and semantic constraints during annotation. We use this data to develop graph-prediction models, finding them to be good at entity identification and local relation extraction, while our corpus facilitates further exploration of challenging long-range relations.

cs.CL↗
arXiv2020

Text Recognition in Real Scenarios with a Few Labeled Samples

Scene text recognition (STR) is still a hot research topic in computer vision field due to its various applications. Existing works mainly focus on learning a general model with a huge number of synthetic text images to recognize unconstrained scene texts, and have achieved substantial progress. However, these methods are not quite applicable in many real-world scenarios where 1) high recognition accuracy is required, while 2) labeled samples are lacked. To tackle this challenging problem, this paper proposes a few-shot adversarial sequence domain adaptation (FASDA) approach to build sequence adaptation between the synthetic source domain (with many synthetic labeled samples) and a specific target domain (with only some or a few real labeled samples). This is done by simultaneously learning each character's feature representation with an attention mechanism and establishing the corresponding character-level latent subspace with adversarial learning. Our approach can maximize the character-level confusion between the source domain and the target domain, thus achieves the sequence-level adaptation with even a small number of labeled samples in the target domain. Extensive experiments on various datasets show that our method significantly outperforms the finetuning scheme, and obtains comparable performance to the state-of-the-art STR methods.

cs.CV↗
arXiv2019

Structured Minimally Supervised Learning for Neural Relation Extraction

We present an approach to minimally supervised relation extraction that combines the benefits of learned representations and structured learning, and accurately predicts sentence-level relation mentions given only proposition-level supervision from a KB. By explicitly reasoning about missing data during learning, our approach enables large-scale training of 1D convolutional neural networks while mitigating the issue of label noise inherent in distant supervision. Our approach achieves state-of-the-art results on minimally supervised sentential relation extraction, outperforming a number of baselines, including a competitive approach that uses the attention layer of a purely neural model.

cs.CL↗
arXiv2019

Spatial clustering and common regulatory elements correlate with coordinated gene expression

Many cellular responses to surrounding cues require temporally concerted transcriptional regulation of multiple genes. In prokaryotic cells, a single-input-module motif with one transcription factor regulating multiple target genes can generate coordinated gene expression. In eukaryotic cells, transcriptional activity of a gene is affected by not only transcription factors but also the epigenetic modifications and three-dimensional chromosome structure of the gene. To examine how local gene environment and transcription factor regulation are coupled, we performed a combined analysis of time-course RNA-seq data of TGF-\b{eta} treated MCF10A cells and related epigenomic and Hi-C data. Using Dynamic Regulatory Events Miner (DREM), we clustered differentially expressed genes based on gene expression profiles and associated transcription factors. Genes in each class have similar temporal gene expression patterns and share common transcription factors. Next, we defined a set of linear and radial distribution functions, as used in statistical physics, to measure the distributions of genes within a class both spatially and linearly along the genomic sequence. Remarkably, genes within the same class despite sometimes being separated by tens of million bases (Mb) along genomic sequence show a significantly higher tendency to be spatially close despite sometimes being separated by tens of Mb along the genomic sequence than those belonging to different classes do. Analyses extended to the process of mouse nervous system development arrived at similar conclusions. Future studies will be able to test whether this spatial organization of chromosomes contributes to concerted gene expression.

q-bio.SC↗
arXiv2018

Edit Probability for Scene Text Recognition

We consider the scene text recognition problem under the attention-based encoder-decoder framework, which is the state of the art. The existing methods usually employ a frame-wise maximal likelihood loss to optimize the models. When we train the model, the misalignment between the ground truth strings and the attention's output sequences of probability distribution, which is caused by missing or superfluous characters, will confuse and mislead the training process, and consequently make the training costly and degrade the recognition accuracy. To handle this problem, we propose a novel method called edit probability (EP) for scene text recognition. EP tries to effectively estimate the probability of generating a string from the output sequence of probability distribution conditioned on the input image, while considering the possible occurrences of missing/superfluous characters. The advantage lies in that the training process can focus on the missing, superfluous and unrecognized characters, and thus the impact of the misalignment problem can be alleviated or even overcome. We conduct extensive experiments on standard benchmarks, including the IIIT-5K, Street View Text and ICDAR datasets. Experimental results show that the EP can substantially boost scene text recognition performance.

cs.CV↗
arXiv2018

AON: Towards Arbitrarily-Oriented Text Recognition

Recognizing text from natural images is a hot research topic in computer vision due to its various applications. Despite the enduring research of several decades on optical character recognition (OCR), recognizing texts from natural images is still a challenging task. This is because scene texts are often in irregular (e.g. curved, arbitrarily-oriented or seriously distorted) arrangements, which have not yet been well addressed in the literature. Existing methods on text recognition mainly work with regular (horizontal and frontal) texts and cannot be trivially generalized to handle irregular texts. In this paper, we develop the arbitrary orientation network (AON) to directly capture the deep features of irregular texts, which are combined into an attention-based decoder to generate character sequence. The whole network can be trained end-to-end by using only images and word-level annotations. Extensive experiments on various benchmarks, including the CUTE80, SVT-Perspective, IIIT5k, SVT and ICDAR datasets, show that the proposed AON-based method achieves the-state-of-the-art performance in irregular datasets, and is comparable to major existing methods in regular datasets.

cs.CV↗
arXiv2017

Focusing Attention: Towards Accurate Text Recognition in Natural Images

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state of the art is the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. However, we observe that existing attention-based methods perform poorly on complicated and/or low-quality images. One major reason is that existing methods cannot get accurate alignments between feature areas and targets for such images. We call this phenomenon "attention drift". To tackle this problem, in this paper we propose the FAN (the abbreviation of Focusing Attention Network) method that employs a focusing attention mechanism to automatically draw back the drifted attention. FAN consists of two major components: an attention network (AN) that is responsible for recognizing character targets as in the existing methods, and a focusing network (FN) that is responsible for adjusting attention by evaluating whether AN pays attention properly on the target areas in the images. Furthermore, different from the existing methods, we adopt a ResNet-based network to enrich deep representations of scene text images. Extensive experiments on various benchmarks, including the IIIT5k, SVT and ICDAR datasets, show that the FAN method substantially outperforms the existing methods.

cs.CV↗
arXiv2017

Label Propagation on K-partite Graphs with Heterophily

In this paper, for the first time, we study label propagation in heterogeneous graphs under heterophily assumption. Homophily label propagation (i.e., two connected nodes share similar labels) in homogeneous graph (with same types of vertices and relations) has been extensively studied before. Unfortunately, real-life networks are heterogeneous, they contain different types of vertices (e.g., users, images, texts) and relations (e.g., friendships, co-tagging) and allow for each node to propagate both the same and opposite copy of labels to its neighbors. We propose a $\mathcal{K}$-partite label propagation model to handle the mystifying combination of heterogeneous nodes/relations and heterophily propagation. With this model, we develop a novel label inference algorithm framework with update rules in near-linear time complexity. Since real networks change over time, we devise an incremental approach, which supports fast updates for both new data and evidence (e.g., ground truth labels) with guaranteed efficiency. We further provide a utility function to automatically determine whether an incremental or a re-modeling approach is favored. Extensive experiments on real datasets have verified the effectiveness and efficiency of our approach, and its superiority over the state-of-the-art label propagation methods.

cs.LG↗
arXiv2016

Online Learning for Wireless Distributed Computing

There has been a growing interest for Wireless Distributed Computing (WDC), which leverages collaborative computing over multiple wireless devices. WDC enables complex applications that a single device cannot support individually. However, the problem of assigning tasks over multiple devices becomes challenging in the dynamic environments encountered in real-world settings, considering that the resource availability and channel conditions change over time in unpredictable ways due to mobility and other factors. In this paper, we formulate a task assignment problem as an online learning problem using an adversarial multi-armed bandit framework. We propose MABSTA, a novel online learning algorithm that learns the performance of unknown devices and channel qualities continually through exploratory probing and makes task assignment decisions by exploiting the gained knowledge. For maximal adaptability, MABSTA is designed to make no stochastic assumption about the environment. We analyze it mathematically and provide a worst-case performance guarantee for any dynamic environment. We also compare it with the optimal offline policy as well as other baselines via emulations on trace-data obtained from a wireless IoT testbed, and show that it offers competitive and robust performance in all cases. To the best of our knowledge, MABSTA is the first online algorithm in this domain of task assignment problems and provides provable performance guarantee.

cs.LG↗
arXiv2013

Coupling between switching regulation and torque generation in bacterial flagellar motor

The bacterial flagellar motor plays a crucial role in both bacterial locomotion and chemotaxis. Recent experiments reveal that the switching dynamics of the motor depends on the motor rotation speed, and thus the motor torque, non-monotonically. Here we present a unified mathematical model which models motor torque generation based on experimental torque-speed curves and torque-dependent switching based on the conformational spread model. The model successfully reproduces the observed switching rate as a function of the rotation speed, and provides a generic physical explanation independent of most details. A stator affects the switching dynamics through two mechanisms: accelerating the conformation flipping rates of individual rotor switching units, which favours slower motor speed and thus increasing torque; and affecting more switching units within unit time, which favours faster speed. Consequently, the switching rate shows a maximum at intermediate speed. Our model predicts that a motor switches more often with more stators. The load-switching relation may serve as a mechanism for sensing the physical environment, similar to the chemotaxis system for sensing the chemical environment. It may also coordinate the switch dynamics of motors within a cell.

physics.bio-ph↗
arXiv2010

Steps and bumps: precision extraction of discrete states of molecular machines using physically-based, high-throughput time series analysis

We report new statistical time-series analysis tools providing significant improvements in the rapid, precision extraction of discrete state dynamics from large databases of experimental observations of molecular machines. By building physical knowledge and statistical innovations into analysis tools, we demonstrate new techniques for recovering discrete state transitions buried in highly correlated molecular noise. We demonstrate the effectiveness of our approach on simulated and real examples of step-like rotation of the bacterial flagellar motor and the F1-ATPase enzyme. We show that our method can clearly identify molecular steps, symmetries and cascaded processes that are too weak for existing algorithms to detect, and can do so much faster than existing algorithms. Our techniques represent a major advance in the drive towards automated, precision, highthroughput studies of molecular machine dynamics. Modular, open-source software that implements these techniques is provided at http://www.eng.ox.ac.uk/samp/members/max/software/

q-bio.QM↗
arXiv2002

Temperature dependence of optical rotation study on parity-violating phase transition of D-, L-, and DL-alanine

Chiral molecules are characterized by a specific optical rotation angle. An experimental method was presented to dissect the temperature dependence of the optical rotation angle with the molecular chirality of D-alanine, L-alanine and DL-alanine crystals. Salam hypothesis predicted that quantum mechanical cooperative and condensation phenomena may give rise to a second order phase transition below a critical temperature linking the transformation of D-amino acids to L-amino acids due to parity-violating energy difference. The temperature- dependent measurement of the optical rotation angle of D-, L- and DL-alanine crystals provided the direct evidence of the phase transition, but denied the configuration change from D-alanine to L-alanine. New views on Salam hypothesis are presented to demonstrate its importance in the application of low temperature enantiomeric separation and the origin of biochirality.

physics.bio-ph↗
arXiv2002

A Review of Salam Phase Transition in Protein Amino Acids: Implication for Biomolecular Homochirality

The origin of chirality, closely related to the evolution of life on the earth, has long been debated. In 1991, Abdus Salam suggested a novel approach to achieve biomolecular homochirality by a phase transition. In his subsequent publication, he predicted that this phase transition could eventually change D-amino acids to L-amino acids as C -H bond would break and H atom became a superconductive atom. Since many experiments denied the configuration change in amino acids, Salam hypothesis aroused suspicion. This paper is aimed to provide direct experimental evidence of a phase transition in alanine, valine single crystals but deny the configuration change of D- to L- enantiomers. New views on Salam phase transition are presented to revalidate its great importance in the origin of homochirality.

physics.bio-ph↗
arXiv2002

Low Temperature Dependence in 1H CRAMPS & 13C CP/MAS ssNMR Spectra of Alanine Enantiomers

1H RAMPS solid state NMR and 13C CP/MAS spectra of D- and L- alanine crystals were measured at temperature ranging from 220K to 295K. Some major points are as follows: (1)alpha-H and protons in the methyl group showed a spin-spin coupling effect as temperature went down in the 1H- MAS spectra of D-alanine. (2) In the 1H CRAMPS spectra, peak widths of alpha-H and beta-H showed a different trend between alanine enantiomers when temperature varied.(3) The dynamic behavior of 13C spectra showed an obvious change around 250K, which was supposed as a phase transition in alanine crystal. In the discussion section, we give these experimental phenomena a possible explanation related to chirality.

physics.bio-ph↗
arXiv2002

Search for the Parity-Violating Effects between D- and L-alanine

The contribution of parity-violating effects to the phase transition of the D-/L-alanine crystals was confirmed by 1H CRAMPS solid state NMR, DC-magnetic susceptibilities and ultrasonic measurements. It was found that the spin relaxation mechanism of alpha-H nucleus of D-alanine molecule is different from L-alanine and the effect is stronger than that of L-alanine. In addition, D-alanine undergoes a magnetic phase transition at a field of 1.0T, which is confirmed by a peak-form ultrasonic attenuation curve. DC-magnetic susceptibilities measurements of L-alanine also indicate abnormal magnetic properties, which is accompanied by a step-form ultrasonic attenuation jump and its mechanism seems different from that of D-alanine. The phase transition is considered to act as a cooperating amplification mechanism of the P-odd effects at the molecular level.

physics.chem-ph↗