arXiv subjects

Jingchen Feng

Publications and source records attributed to Jingchen Feng.

11 recordsLinked to original sources

arXiv2024

Criticality enhances the reinforcement of disordered networks by rigid inclusions

The mechanical properties of biological materials are spatially heterogeneous. Typical tissues are made up of a spanning fibrous extracellular matrix in which various inclusions, such as living cells, are embedded. While the influence of inclusions on the stiffness of common elastic materials such as rubber has been studied for decades and can be understood in terms of the volume fraction and shape of inclusions, the same is not true for disordered filamentous and fibrous networks. Recent work has shown that, in isolation, such networks exhibit unusual viscoelastic behavior indicative of an underlying mechanical phase transition controlled by network connectivity and strain. How this behavior is modified when inclusions are present is unclear. Here, we present a theoretical and computational study of the influence of rigid inclusions on the mechanics of disordered elastic networks near the connectivity-controlled central force rigidity transition. Combining scaling theory and coarse-grained simulations, we predict and confirm an anomalously strong dependence of the composite stiffness on inclusion volume fraction, beyond that seen in ordinary composites. This stiffening exceeds the well-established volume fraction-dependent stiffening expected in conventional composites, e.g., as an elastic analogue of the classic volume fraction dependent increase in the viscosity of liquids first identified by Einstein. We show that this enhancement is a consequence of the interplay between inter-particle spacing and an emergent correlation length, leading to an effective finite-size scaling imposed by the presence of inclusions. We outline the expected scaling of the shear modulus and strain fluctuations with the inclusion volume fraction and network connectivity, confirm these predictions in simulations, and discuss potential experimental tests and implications for our predictions in real systems.

cond-mat.soft↗

arXiv2023

Learning from Negative User Feedback and Measuring Responsiveness for Sequential Recommenders

Sequential recommenders have been widely used in industry due to their strength in modeling user preferences. While these models excel at learning a user's positive interests, less attention has been paid to learning from negative user feedback. Negative user feedback is an important lever of user control, and comes with an expectation that recommenders should respond quickly and reduce similar recommendations to the user. However, negative feedback signals are often ignored in the training objective of sequential retrieval models, which primarily aim at predicting positive user interactions. In this work, we incorporate explicit and implicit negative user feedback into the training objective of sequential recommenders in the retrieval stage using a "not-to-recommend" loss function that optimizes for the log-likelihood of not recommending items with negative feedback. We demonstrate the effectiveness of this approach using live experiments on a large-scale industrial recommender system. Furthermore, we address a challenge in measuring recommender responsiveness to negative feedback by developing a counterfactual simulation framework to compare recommender responses between different user actions, showing improved responsiveness from the modeling change.

cs.IR↗

arXiv2022

Recency Dropout for Recurrent Recommender Systems

Recurrent recommender systems have been successful in capturing the temporal dynamics in users' activity trajectories. However, recurrent neural networks (RNNs) are known to have difficulty learning long-term dependencies. As a consequence, RNN-based recommender systems tend to overly focus on short-term user interests. This is referred to as the recency bias, which could negatively affect the long-term user experience as well as the health of the ecosystem. In this paper, we introduce the recency dropout technique, a simple yet effective data augmentation technique to alleviate the recency bias in recurrent recommender systems. We demonstrate the effectiveness of recency dropout in various experimental settings including a simulation study, offline experiments, as well as live experiments on a large-scale industrial recommendation platform.

cs.IR↗

arXiv2021

ShopTalk: A System for Conversational Faceted Search

We present ShopTalk, a multi-turn conversational faceted search system for shopping that is designed to handle large and complex schemas that are beyond the scope of state of the art slot-filling systems. ShopTalk decouples dialog management from fulfillment, thereby allowing the dialog understanding system to be domain-agnostic and not tied to the particular shopping application. The dialog understanding system consists of a deep-learned Contextual Language Understanding module, which interprets user utterances, and a primarily rules-based Dialog-State Tracker (DST), which updates the dialog state and formulates search requests intended for the fulfillment engine. The interface between the two modules consists of a minimal set of domain-agnostic "intent operators," which instruct the DST on how to update the dialog state. ShopTalk was deployed in 2020 on the Google Assistant for Shopping searches.

cs.CL↗

arXiv2020

Compression stiffening of fibrous networks with stiff inclusions

Tissues commonly consist of cells embedded within a fibrous biopolymer network. Whereas cell-free reconstituted biopolymer networks typically soften under applied uniaxial compression, various tissues, including liver, brain, and fat, have been observed to instead stiffen when compressed. The mechanism for this compression stiffening effect is not yet clear. Here, we demonstrate that when a material composed of stiff inclusions embedded in a fibrous network is compressed, heterogeneous rearrangement of the inclusions can induce tension within the interstitial network, leading to a macroscopic crossover from an initial bending-dominated softening regime to a stretching-dominated stiffening regime, which occurs before and independently of jamming of the inclusions. Using a coarse-grained particle-network model, we first establish a phase diagram for compression-driven, stretching-dominated stress propagation and jamming in uniaxially compressed 2- and 3-dimensional systems. Then, we demonstrate that a more detailed computational model of stiff inclusions in a subisostatic semiflexible fiber network exhibits quantitative agreement with the predictions of our coarse-grained model as well as qualitative agreement with experiments.

cond-mat.soft↗

arXiv2017

Normal stress anisotropy and marginal stability in athermal elastic networks

Hydrogels of semiflexible biopolymers such as collagen have been shown to contract axially under shear strain, in contrast to the axial dilation observed for most elastic materials. Recent work has shown that this behavior can be understood in terms of the porous, two-component nature and consequent time-dependent compressibility of hydrogels. The apparent normal stress measured by a torsional rheometer reflects only the tensile contribution of the axial component $\sigma_{zz}$ on long (compressible) timescales, crossing over to the first normal stress difference, $N_1 = \sigma_{xx}-\sigma_{zz}$ at short (incompressible) times. While the behavior of $N_1$ is well understood for isotropic viscoelastic materials undergoing affine shear deformation, biopolymer networks are often anisotropic and deform nonaffinely. Here, we numerically study the normal stresses that arise under shear in subisostatic, athermal semiflexible polymer networks. We show that such systems exhibit strong deviations from affine behavior and that these anomalies are controlled by a rigidity transition as a function of strain.

cond-mat.soft↗

arXiv2017

Hindrances to precise recovery of cellular forces in fibrous biopolymer networks

How cells move through the three-dimensional extracellular matrix (ECM) is of increasing interest in attempts to understand important biological processes such as cancer metastasis. Just as in motion on flat surfaces, it is expected that experimental measurements of cell-generated forces will provide valuable information for uncovering the mechanisms of cell migration. However, the recovery of forces in fibrous biopolymer networks may suffer from large errors. Here, within the framework of lattice-based models, we explore possible issues in force recovery by solving the inverse problem: how can one determine the forces cells exert to their surroundings from the deformation of the ECM? Our results indicate that irregular cell traction patterns, the uncertainty of local fiber stiffness, the non-affine nature of ECM deformations and inadequate knowledge of network topology will all prevent the precise force determination. At the end, we discuss possible ways of overcoming these difficulties.

physics.bio-ph↗

arXiv2017

Phenomenlogical Modeling of Durotaxis

Cells exhibit qualitatively different behaviors on substrates with different rigidities. The fact that cells are more polarized on the stiffer substrate motivates us to construct a two-dimensional cell with the distribution of focal adhesions dependent on substrate rigidities. This distribution affects the forces exerted by the cell and thereby determines its motion. Our model reproduces the experimental observation that the persistence time is higher on the stiffer substrate. This stiffness dependent persistence will lead to durotaxis, the preference in moving towards stiffer substrates. This propensity is characterized by the durotaxis index first defined in experiments. We derive and validate a 2D corresponding Fokker-Planck equation associated with our model. Our approach highlights the possible role of the focal adhesion arrangement in durotaxis.

physics.bio-ph↗

arXiv2016

Modeling delayed processes in biological systems

Delayed processes are ubiquitous in biological systems and are often characterized by delay differential equations (DDEs) and their extension to include stochastic effects. DDEs do not explicitly incorporate intermediate states associated with a delayed process but instead use an estimated average delay time. In an effort to examine the validity of this approach, we study systems with significant delays by explicitly incorporating intermediate steps. We show by that such explicit models often yield significantly different equilibrium distributions and transition times as compared to DDEs with deterministic delay values. Additionally, different explicit models with qualitatively different dynamics can give rise to the same DDEs revealing important ambiguities. We also show that DDE-based predictions of oscillatory behavior may fail for the corresponding explicit model.

q-bio.QM↗

arXiv2015

Nonlinear elasticity of disordered fiber networks

Disordered biopolymer gels have striking mechanical properties including strong nonlinearities. In the case of athermal gels (such as collagen-I) the nonlinearity has long been associated with a crossover from a bending dominated to a stretching dominated regime of elasticity. The physics of this crossover is related to the existence of a central-force isostatic point and to the fact that for most gels the bending modulus is small. This crossover induces scaling behavior for the elastic moduli. In particular, for linear elasticity such a scaling law has been demonstrated [Broedersz et al. Nature Physics, 2011 7, 983]. In this work we generalize the scaling to the nonlinear regime with a two-parameter scaling law involving three critical exponents. We test the scaling law numerically for two disordered lattice models, and find a good scaling collapse for the shear modulus in both the linear and nonlinear regimes. We compute all the critical exponents for the two lattice models and discuss the applicability of our results to real systems.

cond-mat.soft↗

arXiv2014

Alignment and Nonlinear Elasticity in Biopolymer Gels

We present a Landau type theory for the non-linear elasticity of biopolymer gels with a part of the order parameter describing induced nematic order of fibers in the gel. We attribute the non-linear elastic behavior of these materials to fiber alignment induced by strain. We suggest an application to contact guidance of cell motility in tissue. We compare our theory to simulation of a disordered lattice model for biopolymers. We treat homogeneous deformations such as simple shear, hydrostatic expansion, and simple extension, and obtain good agreement between theory and simulation. We also consider a localized perturbation which is a simple model for a contracting cell in a medium.

cond-mat.soft↗