Search Results for author:
Found 110 papers, 70 papers with code
What’s Hidden in a One-layer Randomly Weighted Transformer?
Hidden within a one-layer randomly weighted Transformer, we find that subnetworks that can achieve 29. 45/17. 29 BLEU on IWSLT14/WMT14.
Train Big, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers
Since hardware resources are limited, the objective of training deep learning models is typically to maximize accuracy subject to the time and memory constraints of training and inference.
The ArtBench Dataset: Benchmarking Generative Models with Artworks
We introduce ArtBench-10, the first class-balanced, high-quality, cleanly annotated, and standardized dataset for benchmarking artwork generation.
Domain-Adaptive Text Classification with Structured Knowledge from Unlabeled Data
Then during training, DASK injects pivot-related knowledge graph information into source domain texts.
Squeezeformer: An Efficient Transformer for Automatic Speech Recognition
After reexamining the design choices for both the macro and micro-architecture of Conformer, we propose the Squeezeformer model, which consistently outperforms the state-of-the-art ASR models under the same training schemes.
Ranked #26 on
Speech Recognition
on LibriSpeech test-clean
UnrealNAS: Can We Search Neural Architectures with Unreal Data?
Neural architecture search (NAS) has shown great success in the automatic design of deep neural networks (DNNs).
Cross-Domain Object Detection with Mean-Teacher Transformer
With these strategies, more accurate pseudo labels can be obtained, and knowledge can be better transferred from source to target, thus improving the cross-domain capability of the detection transformer.
PreTraM: Self-Supervised Pre-training via Connecting Trajectory and Map
It is hard to replicate these approaches in trajectory forecasting due to the lack of adequate trajectory data (e. g., 34K samples in the nuScenes dataset).
K-LITE: Learning Transferable Visual Models with External Knowledge
In this paper, we propose K-LITE (Knowledge-augmented Language-Image Training and Evaluation), a simple strategy to leverage external knowledge to build transferable visual systems: In training, it enriches entities in natural language with WordNet and Wiktionary knowledge, leading to an efficient and scalable approach to learning image representations that can understand both visual concepts and their knowledge; In evaluation, the natural language is also augmented with external knowledge and then used to reference learned visual concepts (or describe new ones) to enable zero-shot and few-shot transfer of the pre-trained models.
A Fast Post-Training Pruning Framework for Transformers
Pruning is an effective way to reduce the huge inference cost of large Transformer models.
Staged Training for Transformer Language Models
As an alternative, we consider a staged training setup that begins with a small model and incrementally increases the amount of compute used for training by applying a "growth operator" to increase the model depth and width.
NovelD: A Simple yet Effective Exploration Criterion
We analyze NovelD thoroughly in MiniGrid and found that empirically it helps the agent explore the environment more uniformly with a focus on exploring beyond the boundary.
Differentiable NAS Framework and Application to Ads CTR Prediction
Neural architecture search (NAS) methods aim to automatically find the optimal deep neural network (DNN) architecture as measured by a given objective function, typically some combination of task accuracy and inference efficiency.
Click-Through Rate Prediction
Natural Language Processing
+1
Multi-source Few-shot Domain Adaptation
Multi-source Domain Adaptation (MDA) aims to transfer predictive models from multiple, fully-labeled source domains to an unlabeled target domain.
What's Hidden in a One-layer Randomly Weighted Transformer?
Hidden within a one-layer randomly weighted Transformer, we find that subnetworks that can achieve 29. 45/17. 29 BLEU on IWSLT14/WMT14.
Emotion Recognition from Multiple Modalities: Fundamentals and Methodologies
In this tutorial, we discuss several key aspects of multi-modal emotion recognition (MER).
How Much Can CLIP Benefit Vision-and-Language Tasks?
Most existing Vision-and-Language (V&L) models rely on pre-trained visual encoders, using a relatively small set of manually-annotated data (as compared to web-crawled data), to perceive the visual world.
Ranked #5 on
Visual Entailment
on SNLI-VE val
(using extra training data)
Scene-aware Learning Network for Radar Object Detection
In this paper, we propose a scene-aware radar learning framework for accurate and robust object detection.
Learned Token Pruning for Transformers
We extensively test the performance of LTP on GLUE tasks and show that our method outperforms the prior state-of-the-art token pruning methods by up to ~2. 5% higher accuracy with the same amount of FLOPs.
Affective Image Content Analysis: Two Decades Review and New Perspectives
Images can convey rich semantics and induce various emotions in viewers.
Invariant Information Bottleneck for Domain Generalization
IIB significantly outperforms IRM on synthetic datasets, where the pseudo-invariant features and geometric skews occur, showing the effectiveness of proposed formulation in overcoming failure modes of IRM.
Image2Point: 3D Point-Cloud Understanding with 2D Image Pretrained Models
We discover that we can indeed use the same architecture and pretrained weights of a neural net model to understand both images and point-clouds.
LEAP: Learnable Pruning for Transformer-based Models
Moreover, in order to reduce hyperparameter tuning, a novel adaptive regularization coefficient is deployed to control the regularization penalty adaptively.
HAO: Hardware-aware neural Architecture Optimization for Efficient Inference
Automatic algorithm-hardware co-design for DNN has shown great success in improving the performance of DNNs on FPGAs.
Integer-only Zero-shot Quantization for Efficient Speech Recognition
End-to-end neural network models achieve improved performance on various automatic speech recognition (ASR) tasks.
Prototypical Cross-domain Self-supervised Learning for Few-shot Unsupervised Domain Adaptation
In this paper, we propose an end-to-end Prototypical Cross-domain Self-Supervised Learning (PCS) framework for Few-shot Unsupervised Domain Adaptation (FUDA).
Ranked #3 on
Semantic Segmentation
on DensePASS
A Survey of Quantization Methods for Efficient Neural Network Inference
Thus, it is not surprising that quantization has emerged recently as an important and very active sub-area of research in the efficient implementation of computations associated with Neural Networks.
Region Similarity Representation Learning
We present Region Similarity Representation Learning (ReSim), a new approach to self-supervised representation learning for localization-based tasks such as object detection and segmentation.
Self-Supervised Pretraining Improves Self-Supervised Pretraining
Through experimentation on 16 diverse vision datasets, we show HPT converges up to 80x faster, improves accuracy across tasks, and improves the robustness of the self-supervised pretraining process to changes in the image augmentation policy or amount of pretraining data.
Improving Context-Based Meta-Reinforcement Learning with Self-Supervised Trajectory Contrastive Learning
To address this, we propose a novel self-supervised learning task, which we named Trajectory Contrastive Learning (TCL), to improve meta-training.
Hessian-Aware Pruning and Optimal Neural Implant
To address this problem, we introduce a new Hessian Aware Pruning (HAP) method coupled with a Neural Implant approach that uses second-order sensitivity as a metric for structured pruning.
I-BERT: Integer-only BERT Quantization
Transformer based models, like BERT and RoBERTa, have achieved state-of-the-art results in many Natural Language Processing tasks.
Visual Transformers: Where Do Transformers Really Belong in Vision Models?
A recent trend in computer vision is to replace convolutions with transformers.
Reservoir Transformers
We demonstrate that transformers obtain impressive performance even when some of the layers are randomly initialized and never updated.
BeBold: Exploration Beyond the Boundary of Explored Regions
In this paper, we analyze the pros and cons of each method and propose the regulated difference of inverse visitation counts as a simple but effective criterion for IR.
Cross-Domain Sentiment Classification with In-Domain Contrastive Learning
In this paper, we propose a contrastive learning framework for cross-domain sentiment classification.
Annotation-Efficient Untrimmed Video Action Recognition
In this paper, we target a new problem, Annotation-Efficient Video Recognition, to reduce the requirement of annotations for both large amount of samples and the action location.
Emotional Semantics-Preserved and Feature-Aligned CycleGAN for Visual Emotion Adaptation
First, we generate an adapted domain to align the source and target domains on the pixel-level by improving CycleGAN with a multi-scale structured cycle-consistency loss.
FBWave: Efficient and Scalable Neural Vocoders for Streaming Text-To-Speech on the Edge
More efficient variants of FBWave can achieve up to 109x fewer MACs while still delivering acceptable audio quality.
HAWQV3: Dyadic Neural Network Quantization
Current low-precision quantization algorithms often have the hidden cost of conversion back and forth from floating point to quantized integer values.
Curriculum CycleGAN for Textual Sentiment Domain Adaptation with Multiple Sources
C-CycleGAN transfers source samples at instance-level to an intermediate domain that is closer to the target domain with sentiment semantics preserved and without losing discriminative features.
Cross-Domain Sentiment Classification with Contrastive Learning and Mutual Information Maximization
Due to scarcity of labels on the target domain, we introduce mutual information maximization (MIM) apart from CL to exploit the features that best support the final prediction.
Multi-Agent Collaboration via Reward Attribution Decomposition
In this work, we propose Collaborative Q-learning (CollaQ) that achieves state-of-the-art performance in the StarCraft multi-agent challenge and supports ad hoc team play.
Learning Invariant Representations and Risks for Semi-supervised Domain Adaptation
First, we provide a finite sample bound for both classification and regression problems under Semi-DA.
SelfAugment: Automatic Augmentation Policies for Self-Supervised Learning
A common practice in unsupervised representation learning is to use labeled data to evaluate the quality of the learned representations.
ePointDA: An End-to-End Simulation-to-Real Domain Adaptation Framework for LiDAR Point Cloud Segmentation
They require prior knowledge of real-world statistics and ignore the pixel-level dropout noise gap and the spatial feature gap between different domains.
A Review of Single-Source Deep Unsupervised Visual Domain Adaptation
To cope with limited labeled training data, many have attempted to directly apply models trained on a large-scale labeled source domain to another sparsely labeled or unlabeled target domain.
Emotion-Based End-to-End Matching Between Image and Music in Valence-Arousal Space
In this paper, we study end-to-end matching between image and music based on emotions in the continuous valence-arousal (VA) space.
Boundary thickness and robustness in learning models
Using these observations, we show that noise-augmentation on mixup training further increases boundary thickness, thereby combating vulnerability to various forms of adversarial attacks and OOD transforms.
Rethinking Distributional Matching Based Domain Adaptation
In this paper, in order to devise robust DA algorithms, we first systematically analyze the limitations of DM based methods, and then build new benchmarks with more realistic domain shifts to evaluate the well-accepted DM methods.
CoDeNet: Efficient Deployment of Input-Adaptive Object Detection on Embedded FPGAs
Deploying deep learning models on embedded systems has been challenging due to limited computing resources.
Visual Transformers: Token-based Image Representation and Processing for Computer Vision
In this work, we challenge this paradigm by (a) representing images as semantic visual tokens and (b) running transformers to densely model token relationships.
ADAHESSIAN: An Adaptive Second Order Optimizer for Machine Learning
We introduce ADAHESSIAN, a second order stochastic optimization algorithm which dynamically incorporates the curvature of the loss function via ADAptive estimates of the HESSIAN.
SqueezeSegV3: Spatially-Adaptive Convolution for Efficient Point-Cloud Segmentation
Using standard convolutions to process such LiDAR images is problematic, as convolution filters pick up local features that are only active in specific regions in the image.
Ranked #14 on
3D Semantic Segmentation
on SemanticKITTI
PowerNorm: Rethinking Batch Normalization in Transformers
To address this, we propose Power Normalization (PN), a novel normalization scheme that resolves this issue by (i) relaxing zero-mean normalization in BN, (ii) incorporating a running quadratic mean instead of per batch statistics to stabilize fluctuations, and (iii) using an approximate backpropagation for incorporating the running statistics in the forward pass.
Multi-source Domain Adaptation in the Deep Learning Era: A Systematic Survey
Therefore, transferring the learned knowledge from a separate, labeled source domain to an unlabeled or sparsely labeled target domain becomes an appealing alternative.
Train Large, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers
Since hardware resources are limited, the objective of training deep learning models is typically to maximize accuracy subject to the time and memory constraints of training and inference.
Algorithm-hardware Co-design for Deformable Convolution
In this work, we first investigate the overhead of the deformable convolution on embedded FPGA SoCs, and then show the accuracy-latency tradeoffs for a set of algorithm modifications including full versus depthwise, fixed-shape, and limited-range.
MADAN: Multi-source Adversarial Domain Aggregation Network for Domain Adaptation
Finally, feature-level alignment is performed between the aggregated domain and the target domain while training the task network.
An End-to-End Visual-Audio Attention Network for Emotion Recognition in User-Generated Videos
Emotion recognition in user-generated videos plays an important role in human-centered computing.
SqueezeWave: Extremely Lightweight Vocoders for On-device Speech Synthesis
Automatic speech synthesis is a challenging task that is becoming increasingly important as edge devices begin to interact with users through speech.
Sound Audio and Speech Processing
ZeroQ: A Novel Zero Shot Quantization Framework
Importantly, ZeroQ has a very low computational overhead, and it can finish the entire quantization process in less than 30s (0. 5\% of one epoch training time of ResNet50 on ImageNet).
Ranked #1 on
Data Free Quantization
on CIFAR10
(CIFAR-10 W8A8 Top-1 Accuracy metric)
PyHessian: Neural Networks Through the Lens of the Hessian
To illustrate this, we analyze the effect of residual connections and Batch Normalization layers on the trainability of neural networks.
ANODEV2: A Coupled Neural ODE Framework
It has been observed that residual networks can be viewed as the explicit Euler discretization of an Ordinary Differential Equation (ODE).
Domain-Aware Dynamic Networks
In this work, we propose a method to improve the model capacity without increasing inference-time complexity.
Multi-source Distilling Domain Adaptation
Deep neural networks suffer from performance decay when there is domain shift between the labeled source domain and unlabeled target domain, which motivates the research on domain adaptation (DA).
Domain Adaptation
Multi-Source Unsupervised Domain Adaptation
HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networks
However, the search space for a mixed-precision quantization is exponential in the number of layers.
Multi-source Domain Adaptation for Semantic Segmentation
In this paper, we propose to investigate multi-source domain adaptation for semantic segmentation.
Checkmate: Breaking the Memory Wall with Optimal Tensor Rematerialization
We formalize the problem of trading-off DNN training time and memory requirements as the tensor rematerialization optimization problem, a generalization of prior checkpointing strategies.
Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT
In particular, we propose a new group-wise quantization scheme, and we use a Hessian based mix-precision method to compress the model further.
PDANet: Polarity-consistent Deep Attention Network for Fine-grained Visual Emotion Regression
By optimizing the PCR loss, PDANet can generate a polarity preserved attention map and thus improve the emotion regression performance.
Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization without Accessing Target Domain Data
To this end, we propose a new approach of domain randomization and pyramid consistency to learn a model with high generalizability.
ANODEV2: A Coupled Neural ODE Evolution Framework
It has been observed that residual networks can be viewed as the explicit Euler discretization of an Ordinary Differential Equation (ODE).
HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precision
Another challenge is a similar factorial complexity for determining block-wise fine-tuning order when quantizing the model to a target precision.
LATTE: Accelerating LiDAR Point Cloud Annotation via Sensor Fusion, One-Click Annotation, and Tracking
2) One-click annotation: Instead of drawing 3D bounding boxes or point-wise labels, we simplify the annotation to just one click on the target object, and automatically generate the bounding box for the target.
Large Batch Optimization for Deep Learning: Training BERT in 76 minutes
In this paper, we first study a principled layerwise adaptation strategy to accelerate training of deep neural networks using large mini-batches.
Ranked #11 on
Question Answering
on SQuAD1.1 dev
(F1 metric)
Inefficiency of K-FAC for Large Batch Size Training
In stochastic optimization, using large batch sizes during training can leverage parallel resources to produce faster wall-clock training times per training epoch.
ANODE: Unconditionally Accurate Memory-Efficient Gradients for Neural ODEs
ANODE has a memory footprint of O(L) + O(N_t), with the same computational cost as reversing ODE solve.
Clustering Multivariate Time Series
Multivariate Time Series Imputation
Large-Batch Training for LSTM and Beyond
LEGW enables Sqrt Scaling scheme to be useful in practice and as a result we achieve much better results than the Linear Scaling learning rate scheme.
Trust Region Based Adversarial Attack on Neural Networks
To address this problem, we present a new family of trust region based adversarial attacks, with the goal of computing adversarial perturbations efficiently.
FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search
Due to this, previous neural architecture search (NAS) methods are computationally expensive.
Ranked #558 on
Image Classification
on ImageNet
Parameter Re-Initialization through Cyclical Batch Size Schedules
We demonstrate the ability of our method to improve language modeling performance by up to 7. 91 perplexity and reduce training iterations by up to $61\%$, in addition to its flexibility in enabling snapshot ensembling and use with adversarial training.
Ranked #48 on
Natural Language Inference
on SNLI
Mixed Precision Quantization of ConvNets via Differentiable Neural Architecture Search
Recent work in network quantization has substantially reduced the time and space complexity of neural network inference, enabling their deployment on embedded and mobile devices with limited computational and memory resources.
Synetgy: Algorithm-hardware Co-design for ConvNet Accelerators on Embedded FPGAs
DiracDeltaNet achieves competitive accuracy on ImageNet (88. 7\% top-5), but with 42$\times$ fewer parameters and 48$\times$ fewer OPs than VGG16.
Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge
This study assesses the state-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i. e., 2012-2018.
A Novel Domain Adaptation Framework for Medical Image Segmentation
Our biophysics based domain adaptation achieves better results, as compared to the existing state-of-the-art GAN model used to create synthetic data for training.
Large batch size training of neural networks with adversarial training and second-order information
Our method exceeds the performance of existing solutions in terms of both accuracy and the number of SGD iterations (up to 1\% and $5\times$, respectively).
SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud
When training our new model on synthetic data using the proposed domain adaptation pipeline, we nearly double test accuracy on real-world data, from 29. 0% to 57. 4%.
Ranked #23 on
3D Semantic Segmentation
on SemanticKITTI
Counterexample-Guided Data Augmentation
We present a novel framework for augmenting data sets for machine learning based on counterexamples.
A LiDAR Point Cloud Generator: from a Virtual World to Autonomous Driving
The framework supports data collection from both auto-driving scenes and user-configured scenes.
Unsupervised Domain Adaptation: from Simulation Engine to the RealWorld
To cope with limited labeled training data, many have attempted to directly apply models trained on a large-scale labeled source domain to another sparsely labeled target domain.
SqueezeNext: Hardware-Aware Neural Network Design
One of the main barriers for deploying neural networks on embedded systems has been large memory and power consumption of existing neural networks.
Hessian-based Analysis of Large Batch Training and Robustness to Adversaries
Extensive experiments on multiple networks show that saddle-points are not the cause for generalization gap of large batch size training, and the results consistently show that large batch converges to points with noticeably higher Hessian spectrum.
Integrated Model, Batch and Domain Parallelism in Training Neural Networks
We propose a new integrated method of exploiting model, batch and domain parallelism for the training of deep neural networks (DNNs) on large distributed-memory computers using minibatch stochastic gradient descent (SGD).
Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
Neural networks rely on convolutions to aggregate spatial information.
Regret Minimization for Partially Observable Deep Reinforcement Learning
Deep reinforcement learning algorithms that estimate state and state-action value functions have been shown to be effective in a variety of challenging domains, including learning control strategies from raw image pixels.
SqueezeSeg: Convolutional Neural Nets with Recurrent CRF for Real-Time Road-Object Segmentation from 3D LiDAR Point Cloud
In this paper, we address semantic segmentation of road-objects from 3D LiDAR point clouds.
Ranked #25 on
3D Semantic Segmentation
on SemanticKITTI
Keynote: Small Neural Nets Are Beautiful: Enabling Embedded Systems with Small Deep-Neural-Network Architectures
Over the last five years Deep Neural Nets have offered more accurate solutions to many problems in speech recognition, and computer vision, and these solutions have surpassed a threshold of acceptability for many applications.
ImageNet Training in Minutes
If we can make full use of the supercomputer for DNN training, we should be able to finish the 90-epoch ResNet-50 training in one minute.
SqueezeDet: Unified, Small, Low Power Fully Convolutional Neural Networks for Real-Time Object Detection for Autonomous Driving
In addition to requiring high accuracy to ensure safety, object detection for autonomous driving also requires real-time inference speed to guarantee prompt vehicle control, as well as small model size and energy efficiency to enable embedded system deployment.
A Metaprogramming and Autotuning Framework for Deploying Deep Learning Applications
On Qualcomm GPUs, we show that our framework enables productive development of target-specific optimizations, and achieves reasonable absolute performance.
How to scale distributed deep learning?
Training time on large datasets for deep neural networks is the principal workflow bottleneck in a number of important applications of deep learning, such as object classification and detection in automatic driver assistance systems (ADAS).
Shallow Networks for High-Accuracy Road Object-Detection
The ability to automatically detect other vehicles on the road is vital to the safety of partially-autonomous and fully-autonomous vehicles.
Boda-RTC: Productive Generation of Portable, Efficient Code for Convolutional Neural Networks on Mobile Computing Platforms
Results are presented for a case study of targeting the Qualcomm Snapdragon 820 mobile computing platform for CNN deployment.
SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
(2) Smaller DNNs require less bandwidth to export a new model from the cloud to an autonomous car.
Ranked #14 on
Network Pruning
on ImageNet
Convolutional Monte Carlo Rollouts in Go
In this work, we present a MCTS-based Go-playing program which uses convolutional networks in all parts.
FireCaffe: near-linear acceleration of deep neural network training on compute clusters
Therefore, the key consideration here is to reduce communication overhead wherever possible, while not degrading the accuracy of the DNN models that we train.
Ranked #601 on
Image Classification
on ImageNet
DeepLogo: Hitting Logo Recognition with the Deep Neural Network Hammer
Recently, there has been a flurry of industrial activity around logo recognition, such as Ditto's service for marketers to track their brands in user-generated images, and LogoGrab's mobile app platform for logo recognition.
Ranked #2 on
Object Detection
on FlickrLogos-32
libHOG: Energy-Efficient Histogram of Oriented Gradient Computation
Histogram of Oriented Gradients (HOG) features are the underlying representation in automotive computer vision applications such as collision avoidance and lane keeping.
DenseNet: Implementing Efficient ConvNet Descriptor Pyramids
Convolutional Neural Networks (CNNs) can provide accurate object classification.
