arXiv subjects

Lijun Wu

Publications and source records attributed to Lijun Wu.

At least 109 records · Page 6Linked to original sources

arXiv2023

Verwey transition as evolution from electronic nematicity to trimerons via electron-phonon coupling

Understanding the driving mechanisms behind metal-insulator transitions (MITs) is a critical step towards controlling material's properties. Since the proposal of charge-order-induced MIT in magnetite Fe3O4 in 1939 by Verwey, the nature of the charge order and its role in the transition have remained elusive-a longstanding challenge in the studies of complex oxides. Recently, a trimeron order was discovered in the low-temperature monoclinic structure of Fe3O4; however, the expected transition entropy change in forming trimeron at the Verwey transition is greater than the observed value, which arises a reexamination of the ground state in the high-temperature phase. Here we use electron diffraction to unveil that a nematic charge order on particular Fe sites emerges in the high-temperature cubic structure of bulk Fe3O4, and that upon cooling, a competitive intertwining of charge and lattice orders leads to the emergence of the Verwey transition. Moreover, MeV ultrafast electron diffraction (UED) provides a dynamic measure of the strong coupling between photoexcited electrons and the X3 phonon modes. Our findings discover a new type of electronic nematicity in correlated materials and offer novel insights into the Verwey transition mechanism in Fe3O4 via the electron-phonon coupling.

cond-mat.str-el↗

arXiv2023

Filler Word Detection with Hard Category Mining and Inter-Category Focal Loss

Filler words like ``um" or ``uh" are common in spontaneous speech. It is desirable to automatically detect and remove them in recordings, as they affect the fluency, confidence, and professionalism of speech. Previous studies and our preliminary experiments reveal that the biggest challenge in filler word detection is that fillers can be easily confused with other hard categories like ``a" or ``I". In this paper, we propose a novel filler word detection method that effectively addresses this challenge by adding auxiliary categories dynamically and applying an additional inter-category focal loss. The auxiliary categories force the model to explicitly model the confusing words by mining hard categories. In addition, inter-category focal loss adaptively adjusts the penalty weight between ``filler" and ``non-filler" categories to deal with other confusing words left in the ``non-filler" category. Our system achieves the best results, with a huge improvement compared to other methods on the PodcastFillers dataset.

eess.AS↗

arXiv2023

AMOM: Adaptive Masking over Masking for Conditional Masked Language Model

Transformer-based autoregressive (AR) methods have achieved appealing performance for varied sequence-to-sequence generation tasks, e.g., neural machine translation, summarization, and code generation, but suffer from low inference efficiency. To speed up the inference stage, many non-autoregressive (NAR) strategies have been proposed in the past few years. Among them, the conditional masked language model (CMLM) is one of the most versatile frameworks, as it can support many different sequence generation scenarios and achieve very competitive performance on these tasks. In this paper, we further introduce a simple yet effective adaptive masking over masking strategy to enhance the refinement capability of the decoder and make the encoder optimization easier. Experiments on \textbf{3} different tasks (neural machine translation, summarization, and code generation) with \textbf{15} datasets in total confirm that our proposed simple method achieves significant performance improvement over the strong CMLM model. Surprisingly, our proposed model yields state-of-the-art performance on neural machine translation (\textbf{34.62} BLEU on WMT16 EN$\to$RO, \textbf{34.82} BLEU on WMT16 RO$\to$EN, and \textbf{34.84} BLEU on IWSLT De$\to$En) and even better performance than the \textbf{AR} Transformer on \textbf{7} benchmark datasets with at least \textbf{2.2$\times$} speedup. Our code is available at GitHub.

cs.CL↗

arXiv2023

De Novo Molecular Generation via Connection-aware Motif Mining

De novo molecular generation is an essential task for science discovery. Recently, fragment-based deep generative models have attracted much research attention due to their flexibility in generating novel molecules based on existing molecule fragments. However, the motif vocabulary, i.e., the collection of frequent fragments, is usually built upon heuristic rules, which brings difficulties to capturing common substructures from large amounts of molecules. In this work, we propose a new method, MiCaM, to generate molecules based on mined connection-aware motifs. Specifically, it leverages a data-driven algorithm to automatically discover motifs from a molecule library by iteratively merging subgraphs based on their frequency. The obtained motif vocabulary consists of not only molecular motifs (i.e., the frequent fragments), but also their connection information, indicating how the motifs are connected with each other. Based on the mined connection-aware motifs, MiCaM builds a connection-aware generator, which simultaneously picks up motifs and determines how they are connected. We test our method on distribution-learning benchmarks (i.e., generating novel molecules to resemble the distribution of a given training set) and goal-directed benchmarks (i.e., generating molecules with target properties), and achieve significant improvements over previous fragment-based baselines. Furthermore, we demonstrate that our method can effectively mine domain-specific motifs for different tasks.

cs.LG↗

arXiv2023

Retrieved Sequence Augmentation for Protein Representation Learning

Protein language models have excelled in a variety of tasks, ranging from structure prediction to protein engineering. However, proteins are highly diverse in functions and structures, and current state-of-the-art models including the latest version of AlphaFold rely on Multiple Sequence Alignments (MSA) to feed in the evolutionary knowledge. Despite their success, heavy computational overheads, as well as the de novo and orphan proteins remain great challenges in protein representation learning. In this work, we show that MSAaugmented models inherently belong to retrievalaugmented methods. Motivated by this finding, we introduce Retrieved Sequence Augmentation(RSA) for protein representation learning without additional alignment or pre-processing. RSA links query protein sequences to a set of sequences with similar structures or properties in the database and combines these sequences for downstream prediction. We show that protein language models benefit from the retrieval enhancement on both structure prediction and property prediction tasks, with a 5% improvement on MSA Transformer on average while being 373 times faster. In addition, we show that our model can transfer to new protein domains better and outperforms MSA Transformer on de novo protein prediction. Our study fills a much-encountered gap in protein prediction and brings us a step closer to demystifying the domain knowledge needed to understand protein sequences. Code is available on https://github.com/HKUNLP/RSA.

q-bio.BM↗

arXiv2022

Direct Molecular Conformation Generation

Molecular conformation generation aims to generate three-dimensional coordinates of all the atoms in a molecule and is an important task in bioinformatics and pharmacology. Previous methods usually first predict the interatomic distances, the gradients of interatomic distances or the local structures (e.g., torsion angles) of a molecule, and then reconstruct its 3D conformation. How to directly generate the conformation without the above intermediate values is not fully explored. In this work, we propose a method that directly predicts the coordinates of atoms: (1) the loss function is invariant to roto-translation of coordinates and permutation of symmetric atoms; (2) the newly proposed model adaptively aggregates the bond and atom information and iteratively refines the coordinates of the generated conformation. Our method achieves the best results on GEOM-QM9 and GEOM-Drugs datasets. Further analysis shows that our generated conformations have closer properties (e.g., HOMO-LUMO gap) with the groundtruth conformations. In addition, our method improves molecular docking by providing better initial conformations. All the results demonstrate the effectiveness of our method and the great potential of the direct approach. The code is released at https://github.com/DirectMolecularConfGen/DMCG

cs.AI↗

arXiv2022

Strongly-overdoped La$_{2-x}$Sr$_x$CuO$_4$: Evidence for Josephson-coupled grains of strongly-correlated superconductor

The interpretation of how superconductivity disappears in cuprates at large hole doping has been controversial. To address this issue, we present an experimental study of single-crystal and thin film samples of La$_{2-x}$Sr$_x$CuO$_4$ (LSCO) with $x\ge0.25$. In particular, measurements of bulk susceptibility on LSCO crystals with $x=0.25$ indicate an onset of diamagnetism at $T_{c1}=38.5$ K, with a sharp transition to a phase with full bulk shielding at $T_{c2}=18$ K, independent of field direction. Strikingly, the in-plane resistivity only goes to zero at $T_{c2}$. Inelastic neutron scattering on $x=0.25$ crystals confirms the presence of low-energy incommensurate magnetic excitations with reduced strength compared to lower doping levels. The ratio of the spin gap to $T_{c2}$ is anomalously large. Our results are consistent with a theoretical prediction for strongly overdoped cuprates by Spivak, Oreto, and Kivelson, in which superconductivity initially develops within disconnected self-organized grains characterized by a reduced hole concentration, with bulk superconductivity occurring only after superconductivity is induced by proximity effect in the surrounding medium of higher hole concentration. Beyond the superconducting-to-metal transition, local differential conductance measurements on an LSCO thin film suggest that regions with pairing correlations survive, but are too dilute to support superconducting order. Future experiments will be needed to test the degree to which these results apply to overdoped cuprates in general.

cond-mat.supr-con↗

arXiv2022

Dual-stage structural response to quenching charge order in magnetite

The Verwey transition in magnetite (Fe3O4 ) is the prototypical metal-insulator transition and has eluded a comprehensive explanation for decades. A major element of the challenge is the complex interplay between charge order and lattice distortions. Here we use ultrafast electron diffraction (UED) to disentangle the roles of charge order and lattice distortions by tracking the transient structural evolution after charge order is melted via ultrafast photoexcitation. A dual stage response is observed in which X3, X1, and Delta5 type structural distortions occur on markedly different timescales of 0.7 to 3.2 ps and longer than 3.2 ps. We propose that these distinct timescales arise because X3 type distortions strongly couple to the trimeron charge order, whereas the Delta5-distortions are more strongly associated with monoclinic to cubic distortions of the overall lattice. Our work aids in clarifying the charge lattice interplay using UED method and illustrates the disentanglement of the complex phases in magnetite.

cond-mat.str-el↗

arXiv2022

Incorporating Pre-training Paradigm for Antibody Sequence-Structure Co-design

Antibodies are versatile proteins that can bind to pathogens and provide effective protection for human body. Recently, deep learning-based computational antibody design has attracted popular attention since it automatically mines the antibody patterns from data that could be complementary to human experiences. However, the computational methods heavily rely on high-quality antibody structure data, which is quite limited. Besides, the complementarity-determining region (CDR), which is the key component of an antibody that determines the specificity and binding affinity, is highly variable and hard to predict. Therefore, the data limitation issue further raises the difficulty of CDR generation for antibodies. Fortunately, there exists a large amount of sequence data of antibodies that can help model the CDR and alleviate the reliance on structure data. By witnessing the success of pre-training models for protein modeling, in this paper, we develop the antibody pre-training language model and incorporate it into the (antigen-specific) antibody design model in a systemic way. Specifically, we first pre-train an antibody language model based on the sequence data, then propose a one-shot way for sequence and structure generation of CDR to avoid the heavy cost and error propagation from an autoregressive manner, and finally leverage the pre-trained antibody model for the antigen-specific antibody generation model with some carefully designed modules. Through various experiments, we show that our method achieves superior performances over previous baselines on different tasks, such as sequence and structure generation and antigen-binding CDR-H3 design.

q-bio.BM↗

arXiv2022

Ultrafast formation of topological defects in a 2D charge density wave

Topological defects play a key role in nonequilibrium phase transitions, ranging from birth of the early universe to quantum critical behavior of ultracold atoms. In solids, transient defects are known to generate a variety of hidden orders not accessible in equilibrium, but how defects are formed at the nanometer lengthscale and femtosecond timescale remains unknown. Here, we employ an intense laser pulse to create topological defects in a 2D charge density wave, and track their morphology and dynamics with ultrafast electron diffraction. Leveraging its high temporal resolution and sensitivity in detecting weak diffuse signals, we discover a dual-stage growth of 1D domain walls within 1 ps, a process not dictated by the order parameter amplitude but instead mediated by a nonthermal population of longitudinal optical phonons. Our work provides a framework for ultrafast engineering of topological defects based on selective excitation of collective modes, opening new avenues for dynamical control of nonequilibrium phases in correlated materials.

cond-mat.mtrl-sci↗

arXiv2022

Improving Temporal Generalization of Pre-trained Language Models with Lexical Semantic Change

Recent research has revealed that neural language models at scale suffer from poor temporal generalization capability, i.e., the language model pre-trained on static data from past years performs worse over time on emerging data. Existing methods mainly perform continual training to mitigate such a misalignment. While effective to some extent but is far from being addressed on both the language modeling and downstream tasks. In this paper, we empirically observe that temporal generalization is closely affiliated with lexical semantic change, which is one of the essential phenomena of natural languages. Based on this observation, we propose a simple yet effective lexical-level masking strategy to post-train a converged language model. Experiments on two pre-trained language models, two different classification tasks, and four benchmark datasets demonstrate the effectiveness of our proposed method over existing temporal adaptation methods, i.e., continual training with new data. Our code is available at \url{https://github.com/zhaochen0110/LMLM}.

cs.CL↗

arXiv2022

Tailoring Molecules for Protein Pockets: a Transformer-based Generative Solution for Structured-based Drug Design

Structure-based drug design is drawing growing attentions in computer-aided drug discovery. Compared with the virtual screening approach where a pre-defined library of compounds are computationally screened, de novo drug design based on the structure of a target protein can provide novel drug candidates. In this paper, we present a generative solution named TamGent (Target-aware molecule generator with Transformer) that can directly generate candidate drugs from scratch for a given target, overcoming the limits imposed by existing compound libraries. Following the Transformer framework (a state-of-the-art framework in deep learning), we design a variant of Transformer encoder to process 3D geometric information of targets and pre-train the Transformer decoder on 10 million compounds from PubChem for candidate drug generation. Systematical evaluation on candidate compounds generated for targets from DrugBank shows that both binding affinity and drugability are largely improved. TamGent outperforms previous baselines in terms of both effectiveness and efficiency. The method is further verified by generating candidate compounds for the SARS-CoV-2 main protease and the oncogenic mutant KRAS G12C. The results show that our method not only re-discovers previously verified drug molecules , but also generates novel molecules with better docking scores, expanding the compound pool and potentially leading to the discovery of novel drugs.

q-bio.BM↗

arXiv2022

Unified 2D and 3D Pre-Training of Molecular Representations

Molecular representation learning has attracted much attention recently. A molecule can be viewed as a 2D graph with nodes/atoms connected by edges/bonds, and can also be represented by a 3D conformation with 3-dimensional coordinates of all atoms. We note that most previous work handles 2D and 3D information separately, while jointly leveraging these two sources may foster a more informative representation. In this work, we explore this appealing idea and propose a new representation learning method based on a unified 2D and 3D pre-training. Atom coordinates and interatomic distances are encoded and then fused with atomic representations through graph neural networks. The model is pre-trained on three tasks: reconstruction of masked atoms and coordinates, 3D conformation generation conditioned on 2D graph, and 2D graph generation conditioned on 3D conformation. We evaluate our method on 11 downstream molecular property prediction tasks: 7 with 2D information only and 4 with both 2D and 3D information. Our method achieves state-of-the-art results on 10 tasks, and the average improvement on 2D-only tasks is 8.3%. Our method also achieves significant improvement on two 3D conformation generation tasks.

cs.LG↗

arXiv2022

RetroGraph: Retrosynthetic Planning with Graph Search

Retrosynthetic planning, which aims to find a reaction pathway to synthesize a target molecule, plays an important role in chemistry and drug discovery. This task is usually modeled as a search problem. Recently, data-driven methods have attracted many research interests and shown promising results for retrosynthetic planning. We observe that the same intermediate molecules are visited many times in the searching process, and they are usually independently treated in previous tree-based methods (e.g., AND-OR tree search, Monte Carlo tree search). Such redundancies make the search process inefficient. We propose a graph-based search policy that eliminates the redundant explorations of any intermediate molecules. As searching over a graph is more complicated than over a tree, we further adopt a graph neural network to guide the search over graphs. Meanwhile, our method can search a batch of targets together in the graph and remove the inter-target duplication in the tree-based search methods. Experimental results on two datasets demonstrate the effectiveness of our method. Especially on the widely used USPTO benchmark, we improve the search success rate to 99.47%, advancing previous state-of-the-art performance for 2.6 points.

cs.AI↗

arXiv2022

A Mutually Reinforced Framework for Pretrained Sentence Embeddings

The lack of labeled data is a major obstacle to learning high-quality sentence embeddings. Recently, self-supervised contrastive learning (SCL) is regarded as a promising way to address this problem. However, the existing works mainly rely on hand-crafted data annotation heuristics to generate positive training samples, which not only call for domain expertise and laborious tuning, but are also prone to the following unfavorable cases: 1) trivial positives, 2) coarse-grained positives, and 3) false positives. As a result, the self-supervision's quality can be severely limited in reality. In this work, we propose a novel framework InfoCSE to address the above problems. Instead of relying on annotation heuristics defined by humans, it leverages the sentence representation model itself and realizes the following iterative self-supervision process: on one hand, the improvement of sentence representation may contribute to the quality of data annotation; on the other hand, more effective data annotation helps to generate high-quality positive samples, which will further improve the current sentence representation model. In other words, the representation learning and data annotation become mutually reinforced, where a strong self-supervision effect can be derived. Extensive experiments are performed based on three benchmark datasets, where notable improvements can be achieved against the existing SCL-based methods.

cs.CL↗

arXiv2022

AF$_2$: Adaptive Focus Framework for Aerial Imagery Segmentation

As a specific semantic segmentation task, aerial imagery segmentation has been widely employed in high spatial resolution (HSR) remote sensing images understanding. Besides common issues (e.g. large scale variation) faced by general semantic segmentation tasks, aerial imagery segmentation has some unique challenges, the most critical one among which lies in foreground-background imbalance. There have been some recent efforts that attempt to address this issue by proposing sophisticated neural network architectures, since they can be used to extract informative multi-scale feature representations and increase the discrimination of object boundaries. Nevertheless, many of them merely utilize those multi-scale representations in ad-hoc measures but disregard the fact that the semantic meaning of objects with various sizes could be better identified via receptive fields of diverse ranges. In this paper, we propose Adaptive Focus Framework (AF$_2$), which adopts a hierarchical segmentation procedure and focuses on adaptively utilizing multi-scale representations generated by widely adopted neural network architectures. Particularly, a learnable module, called Adaptive Confidence Mechanism (ACM), is proposed to determine which scale of representation should be used for the segmentation of different objects. Comprehensive experiments show that AF$_2$ has significantly improved the accuracy on three widely used aerial benchmarks, as fast as the mainstream method.

cs.CV↗

arXiv2022

Dynamic Relation Discovery and Utilization in Multi-Entity Time Series Forecasting

Time series forecasting plays a key role in a variety of domains. In a lot of real-world scenarios, there exist multiple forecasting entities (e.g. power station in the solar system, stations in the traffic system). A straightforward forecasting solution is to mine the temporal dependency for each individual entity through 1d-CNN, RNN, transformer, etc. This approach overlooks the relations between these entities and, in consequence, loses the opportunity to improve performance using spatial-temporal relation. However, in many real-world scenarios, beside explicit relation, there could exist crucial yet implicit relation between entities. How to discover the useful implicit relation between entities and effectively utilize the relations for each entity under various circumstances is crucial. In order to mine the implicit relation between entities as much as possible and dynamically utilize the relation to improve the forecasting performance, we propose an attentional multi-graph neural network with automatic graph learning (A2GNN) in this work. Particularly, a Gumbel-softmax based auto graph learner is designed to automatically capture the implicit relation among forecasting entities. We further propose an attentional relation learner that enables every entity to dynamically pay attention to its preferred relations. Extensive experiments are conducted on five real-world datasets from three different domains. The results demonstrate the effectiveness of A2GNN beyond several state-of-the-art methods.

cs.LG↗

arXiv2021

Incorporating NODE with Pre-trained Neural Differential Operator for Learning Dynamics

Learning dynamics governed by differential equations is crucial for predicting and controlling the systems in science and engineering. Neural Ordinary Differential Equation (NODE), a deep learning model integrated with differential equations, is popular in learning dynamics recently due to its robustness to irregular samples and its flexibility to high-dimensional input. However, the training of NODE is sensitive to the precision of the numerical solver, which makes the convergence of NODE unstable, especially for ill-conditioned dynamical systems. In this paper, to reduce the reliance on the numerical solver, we propose to enhance the supervised signal in the training of NODE. Specifically, we pre-train a neural differential operator (NDO) to output an estimation of the derivatives to serve as an additional supervised signal. The NDO is pre-trained on a class of basis functions and learns the mapping between the trajectory samples of these functions to their derivatives. To leverage both the trajectory signal and the estimated derivatives from NDO, we propose an algorithm called NDO-NODE, in which the loss function contains two terms: the fitness on the true trajectory samples and the fitness on the estimated derivatives that are outputted by the pre-trained NDO. Experiments on various kinds of dynamics show that our proposed NDO-NODE can consistently improve the forecasting accuracy with one pre-trained NDO. Especially for the stiff ODEs, we observe that NDO-NODE can capture the transitions in the dynamics more accurately compared with other regularization methods.

cs.LG↗