Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Zhaojie JuSuggested from co-authorship
- Thomas S. HuangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works178 from public data
- Conformer: Convolution-augmented Transformer for Speech Recognition4,403
This work proposes the convolution-augmented transformer for speech recognition, named Conformer, which significantly outperforms the previous Transformer and CNN based models achieving state-of-the-art accuracies.
- Generative Image Inpainting with Contextual Attention2,597
This work proposes a new deep generative model-based approach which can not only synthesize novel image structures but also explicitly utilize surrounding image features as references during network training to make better predictions.
- Free-Form Image Inpainting With Gated Convolution2,102
The proposed gated convolution solves the issue of vanilla convolution that treats all input pixels as valid ones, generalizes partial convolution by providing a learnable dynamic feature selection mechanism for each channel at each spatial location across all layers.
- 1,955
- Universally Slimmable Networks and Improved Training Techniques473
This work proposes a systematic approach to train universally slimmable networks (US-Nets), extending slimmables to execute at arbitrary width, and generalizing to networks both with and without batch normalization layers, and opens up the possibility to directly evaluate FLOPs-Accuracy spectrum of network architectures.
- Foreground-Aware Image Inpainting402
This work proposes a foreground-aware image inpainting system that explicitly disentangles structure inference and content completion, and shows that by such disentanglement, the contour completion model predicts reasonable contours of objects, and further substantially improves the performance of image inPainting.
- 360
- BigNAS: Scaling up Neural Architecture Search with Big Single-Stage Models344
The proposed BigNAS, an approach that challenges the conventional wisdom that post-processing of the weights is necessary to get good prediction accuracies, is proposed, able to train a single set of shared weights on ImageNet and use these weights to obtain child models whose sizes range from 200 to 1000 MFLOPs.
- 327
- ContextNet: Improving Convolutional Neural Networks for Automatic Speech Recognition with Global Context306
This paper proposes a simple scaling method that scales the widths of ContextNet that achieves good trade-off between computation and accuracy and demonstrates that on the widely used LibriSpeech benchmark, ContextNet achieves a word error rate of 2.1%/4.6%.
- 301
- Combined scaling for zero-shot transfer learning251
A theoretical framework is developed which shows that larger contrastive batch sizes lead to smaller generalization gaps for image-text models such as BASIC, and sheds light on the benefits of large contrastivebatch sizes.
- Generative Adversarial Networks for Image and Video Synthesis: Algorithms and Applications235
An overview of GANs with a special focus on algorithms and applications for visual synthesis is provided and several important techniques to stabilize GAN training are covered, which has a reputation for being notoriously difficult.
- 217
- BigSSL: Exploring the Frontier of Large-Scale Semi-Supervised Learning for Automatic Speech Recognition212
It is found that the combination of pre-training, self-training and scaling up model size greatly increases data efficiency, even for extremely large tasks with tens of thousands of hours of labeled data.
- Engineered monoculture and co-culture of methylotrophic yeast for de novo production of monacolin J and lovastatin from methanol161
This study provides an alternative method for production of monacolin J and lovastatin and reveals the potential of a methylotrophic yeast to produce complicated pharmaceuticals from methanol.
- Pyramid Attention Network for Image Restoration142
A novel Pyramid Attention module for image restoration, which captures long-range feature correspondences from a multi-scale feature pyramid, and is designed to be able to borrow clean signals from their “clean” correspondences at the coarser levels.
- A Better and Faster end-to-end Model for Streaming ASR136
The Conformer RNN-T with Cascaded Encoders offers a better quality and latency tradeoff for streaming ASR.
- 121
- 115
- FastEmit: Low-Latency Streaming ASR with Sequence-Level Emission Regularization108
This work proposes a sequence-level emission regularization method, named FastEmit, that applies latency regularization directly on per-sequence probability in training transducer models, and does not require any alignment.
- 90
- Cascaded Encoders for Unifying Streaming and Non-Streaming ASR89
This work presents cascaded encoders for building a single E2E ASR model that can operate in both streaming and non-streaming mode simultaneously, and shows that the proposed approach outperforms existing E 2E two-pass models, especially on long-form speech.
- 83
- Organic Compound Synthetic Accessibility Prediction Based on the Graph Attention Mechanism73
This study proposed a data-driven interpretable prediction framework called GASA (Graph Attention-based assessment of Synthetic Accessibility) to evaluate the synthetic accessibility of small molecules by distinguishing compounds to be easy- (ES) or hard-to-synthesize (HS).
- 65
- 63
- 60
- Engineered ethanol-driven biosynthetic system for improving production of acetyl-CoA derived drugs in Crabtree-negative yeast60
This study provides a new paradigm for facilitating the high-yield production of acetyl-CoA derived pharmaceuticals and value-added molecules by constructing a new biosynthetic system based on a Crabtree-negative yeast, which grew exceptionally on ethanol and assimilated ethanol directly in cytosol to acetyl -CoA.
- Network Slimming by Slimmable Networks: Towards One-Shot Architecture Search for Channel Numbers59
A simple and one-shot solution to set channel numbers in a neural network to achieve better accuracy under constrained resources (e.g., FLOPs, latency, memory footprint or model size) is presented.
- Wide-activated Deep Residual Networks based Restoration for BPG-compressed Images49
A simple pipeline to achieve high-quality image compression under very low bit-rate is investigated, which significantly reduces the quantity loss and remove visual artifacts for compressed images.
- Dynamic Sparsity Neural Networks for Automatic Speech Recognition46
This paper presents Dynamic Sparsity Neural Networks (DSNN), a trained DSNN that, once trained, can instantly switch to any predefined sparsity configuration at run-time and can greatly ease the training process and simplify deployment in diverse scenarios with resource constraints.
- Scale-Wise Convolution for Image Restoration46
It is shown that properly modeling scale-invariance into neural networks can bring significant benefits to image restoration performance and the proposed network with scale-wise convolution achieves superior performance in multiple image restoration tasks including image super-resolution, image denoising and image compression artifacts removal.
- 45
- 45
- De Novo Production of Plant 4′-Deoxyflavones Baicalein and Oroxylin A from Ethanol in Crabtree-Negative Yeast43
New strategies for the microbial synthesis of 4'-deoxyflavones and other flavonoids are provided and fine-tuning pathway control considerably reduced byproduct and intermediate accumulation and achieved high-level de novo baicalein and oroxylin A production from ethanol.
- A Discriminative Deep Model With Feature Fusion and Temporal Attention for Human Action Recognition42
A novel discriminative deep model based on 3D-CNN and LSTM for both single-target and interaction action recognition to improve the spatiotemporal processing performance and introduces an improved attention mechanism that focuses on each frame individually by assigning different weights in real-time.
- 40
- Can molecular dynamics simulations improve predictions of protein-ligand binding affinity with machine learning?39
The outcomes of the ML models built using various dimensional descriptors and their combinations reveal that the MD refinement with the optimized protocol can improve the predictive performance on the TAF1-BD2 target with considerable structural flexibility, but not for the less flexible JAK1 and DDR1 targets, when taking docking poses as the initial structure.
- 38
- 38
- 38
- Rational generation of lasso peptides based on biosynthetic gene mutations and site-selective chemical modifications37
The discovery of a new lasso peptide stlassin by gene activation based on a Streptomyces heterologous expression system is reported, demonstrating proof-of-concept for the combined mutational/chemical generation of lasso Peptide libraries to support drug lead development.
- 37
- 36
- Deep Object Detector With Attentional Spatiotemporal LSTM for Space Human–Robot Interaction36
This article proposes an attentional global–local information learning network, one of the first attempts to fully use both types of information between frames, and achieves state-of-the-art performance on challenging benchmarks.
- Kinome-wide polypharmacology profiling of small molecules by multi-task graph isomorphism network approach31
A model called AMGU was developed to predict the inhibitory activities of small molecules against 204 different kinases based on the multi-task Graph Isomorphism Network (MT-GIN) with the auxiliary learning and uncertainty weighting strategy, suggesting that the AMGU model has enhanced generalizability due to its great transfer learning capacity.
- 30
- 29
- 28
- 25
- 24
- 24
- 23
- 23
- 22
- 20
- 20
- 19
- 19
- 18
- 18
- 17
- 16
- 16
- 16
- 14
- 14
- 14
- 13
- 13
- 13
- 13
- 12
- 12
- 12
- 11
- 11
- 10
- 10
- 9
- 9
- 9
- 9
- 8
- 7
- 7
- 7
- 7
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 3
- 3
- 3
- 3
- RCV2023 Challenges: Benchmarking Model Training and Inference for Resource-Constrained Deep Learning3
- 3
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-13. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.