Search

Although Transformer-based deep learning models have been widely used in many natural language processing (NLP) tasks as well as computer vision, they suffer from gigantic model size and long latency. Network pruning can reduce the computational cost and model size. However, existing works mainly focus on irregular(sparse) pruning, which often causes irregular computations and extra indices per remained weight. In this work, we propose a Tensor-core inspired hierarchical model compression method to push the performance limit on modern GPUs. We present two modes of the two-step process. In the first mode, we use the Tensor-core aware block-based weight pruning method to exploit model sparsity in a coarse-grained manner and then use low-rank [33] decomposition to further reduce the weight storage in a fine-grained manner.In the second mode, we first use irregular pruning to achieve a highly sparse model and then apply the Tensor-core aware weight constraint on the sparse model to decompose the sparse matrix to several smaller but Tensor-core friendly sub-matrices. Experiments on Transformer, BERTBASE models show the proposed method outperforms the state-of-the-art.

Image segmentation plays an important role in multimodality imaging, especially in fusion structural images offered by CT, MRI with functional images collected by optical technologies, or other novel imaging technologies. In addition, image segmentation also provides detailed structural description for quantitative visualization of treating light distribution in the human body when incorporated with 3D light transport simulation methods. Here, we first use some preprocessing methods such as wavelet denoising to extract the accurate contours of different tissues such as skull, cerebrospinal fluid (CSF), grey matter (GM), and white matter (WM) on 5 MRI head image datasets. We then realize automatic image segmentation with deep learning by using convolutional neural network. We also introduce parallel computing. Such approaches greatly reduced the processing time compared to manual and semiautomatic segmentation and are of great importance in improving the speed and accuracy as more and more samples are being learned. The segmented data of grey and white matter are counted by computer in volume, which indicates the potential of this segmentation technology in diagnosing cerebral atrophy quantitatively. We demonstrate the great potential of such image processing and deep learning-combined automatic tissue image segmentation in neurology medicine.

Transformer-based deep learning models have increasingly demonstrated high accuracy on many natural language processing (NLP) tasks. In this paper, we propose a compression-compilation co-design framework that can guarantee the identified model to meet both resource and real-time specifications of mobile devices. Our framework applies a compiler-aware neural architecture optimization method (CANAO), which can generate the optimal compressed model that balances both accuracy and latency. We are able to achieve up to 7.8x speedup compared with TensorFlow-Lite with only minor accuracy loss. We present two types of BERT applications on mobile devices: Question Answering (QA) and Text Generation. Both can be executed in real-time with latency as low as 45ms. Videos for demonstrating the framework can be found on https://www.youtube.com/watch?v=_WIRvK_2PZI, arXiv admin note: substantial text overlap with arXiv:2009.06823

Recent research demonstrated the promise of using resistive random access memory (ReRAM) as an emerging technology to perform inherently parallel analog domain in-situ matrix-vector multiplication -- the intensive and key computation in deep neural networks (DNNs). However, hardware failure, such as stuck-at-fault defects, is one of the main concerns that impedes the ReRAM devices to be a feasible solution for real implementations. The existing solutions to address this issue usually require an optimization to be conducted for each individual device, which is impractical for mass-produced products (e.g., IoT devices). In this paper, we rethink the value of weight pruning in ReRAM-based DNN design from the perspective of model fault tolerance. And a differential mapping scheme is proposed to improve the fault tolerance under a high stuck-on fault rate. Our method can tolerate almost an order of magnitude higher failure rate than the traditional two-column method in representative DNN tasks. More importantly, our method does not require extra hardware cost compared to the traditional two-column mapping scheme. The improvement is universal and does not require the optimization process for each individual device., Comment: In Proceedings of the 22nd International Symposium on Quality Electronic Design (ISQED), 2021

With the increasing demand to efficiently deploy DNNs on mobile edge devices, it becomes much more important to reduce unnecessary computation and increase the execution speed. Prior methods towards this goal, including model compression and network architecture search (NAS), are largely performed independently and do not fully consider compiler-level optimizations which is a must-do for mobile acceleration. In this work, we first propose (i) a general category of fine-grained structured pruning applicable to various DNN layers, and (ii) a comprehensive, compiler automatic code generation framework supporting different DNNs and different pruning schemes, which bridge the gap of model compression and NAS. We further propose NPAS, a compiler-aware unified network pruning, and architecture search. To deal with large search space, we propose a meta-modeling procedure based on reinforcement learning with fast evaluation and Bayesian optimization, ensuring the total number of training epochs comparable with representative NAS frameworks. Our framework achieves 6.7ms, 5.9ms, 3.9ms ImageNet inference times with 78.2%, 75% (MobileNet-V3 level), and 71% (MobileNet-V2 level) Top-1 accuracy respectively on an off-the-shelf mobile phone, consistently outperforming prior work., Accepted as an oral paper in the Conference on Computer Vision and Pattern Recognition (CVPR), 2021

Pre-trained large-scale language models have increasingly demonstrated high accuracy on many natural language processing (NLP) tasks. However, the limited weight storage and computational speed on hardware platforms have impeded the popularity of pre-trained models, especially in the era of edge computing. In this work, we propose an efficient transformer-based large-scale language representation using hardware-friendly block structure pruning. We incorporate the reweighted group Lasso into block-structured pruning for optimization. Besides the significantly reduced weight storage and computation, the proposed approach achieves high compression rates. Experimental results on different models (BERT, RoBERTa, and DistilBERT) on the General Language Understanding Evaluation (GLUE) benchmark tasks show that we achieve up to 5.0x with zero or minor accuracy degradation on certain task(s). Our proposed method is also orthogonal to existing compact pre-trained language models such as DistilBERT using knowledge distillation, since a further 1.79x average compression rate can be achieved on top of DistilBERT with zero or minor accuracy degradation. It is suitable to deploy the final compressed model on resource-constrained edge devices., Comment: Accepted to Findings of EMNLP 2020

Recently, acupuncture, traditional to Chinese medicine for dozens of centuries, has been receiving recognition as a well-established and widely-cherished methodology among the newly-discovered diagnoses. It is currently embracing an ever-growing and wide-spreading global popularity. However, in comparison with traditional acupuncture utilizing metallic needles, laser acupuncture nevertheless stands out as a non-invasive, painless approach to diseases, eliminating the risk of infection. Due to the advantages of laser acupuncture, we innovatively developed a portable theranostic device adopting laser acupuncture. In our study, a series of experiments were performed to substantiate as well as improve the robustness and stability of this device for future clinical trials. For instance, we adopted a type of pluggable interface designed to improve the stability of laser diode part. In addition, we measured the dark noise of every channel for data interpretation. The results have quantitatively demonstrated the device's great potentiality and reliability in clinical treatments. In future, the device will be tested in noninvasive theranostic clinic trials to fully test the capabilities of treating various diseases.

Image segmentation plays an important role in medical science. One application is multimodality imaging, especially the fusion of structural imaging with functional imaging, which includes CT, MRI and new types of imaging technology such as optical imaging to obtain functional images. The fusion process require precisely extracted structural information, in order to register the image to it. Here we used image enhancement, morphometry methods to extract the accurate contours of different tissues such as skull, cerebrospinal fluid (CSF), grey matter (GM) and white matter (WM) on 5 fMRI head image datasets. Then we utilized convolutional neural network to realize automatic segmentation of images in deep learning way. Such approach greatly reduced the processing time compared to manual and semi-automatic segmentation and is of great importance in improving speed and accuracy as more and more samples being learned. The contours of the borders of different tissues on all images were accurately extracted and 3D visualized. This can be used in low-level light therapy and optical simulation software such as MCVM. We obtained a precise three-dimensional distribution of brain, which offered doctors and researchers quantitative volume data and detailed morphological characterization for personal precise medicine of Cerebral atrophy/expansion. We hope this technique can bring convenience to visualization medical and personalized medicine.

Image segmentation plays an important role in multimodality imaging, especially in fusion structural images offered by CT, MRI with functional images collected by optical technologies or other novel imaging technologies. Plus, image segmentation also provides detailed structure description for quantitative visualization of treating light distribution in the human body when incorporated with 3D light transport simulation method. Here we used image enhancement, operators, and morphometry methods to extract the accurate contours of different tissues such as skull, cerebrospinal fluid (CSF), grey matter (GM) and white matter (WM) on 5 fMRI head image datasets. Then we utilized convolutional neural network to realize automatic segmentation of images in a deep learning way. We also introduced parallel computing. Such approaches greatly reduced the processing time compared to manual and semi-automatic segmentation and is of great importance in improving speed and accuracy as more and more samples being learned. Our results can be used as a criteria when diagnosing diseases such as cerebral atrophy, which is caused by pathological changes in gray matter or white matter. We demonstrated the great potential of such image processing and deep leaning combined automatic tissue image segmentation in personalized medicine, especially in monitoring, and treatments.