Spiking Neural Networks (SNNs) are promising for enabling low-power event-driven data analytics. The best performing SNNs for image recognition tasks are obtained by converting a trained deep learning Analog Neural Network (ANN) composed of Rectified Linear Unit (ReLU) activation to SNN consisting of Integrate-and-Fire (IF) neurons with ""proper"" firing thresholds. However, this has come at the cost of accuracy loss and higher inference latency due to lack of a notion of time. In this work, we propose an ANN to SNN conversion methodology that uses a time-based coding scheme, named Temporal-Switch-Coding (TSC), and a corresponding TSC spiking neuron model. Each input image pixel is presented using two spikes and the timing between the two spiking instants is proportional to the pixel intensity. The real-valued ReLU activations in ANN are encoded using the spike-times of the TSC neurons in the converted TSC-SNN. At most two memory accesses and two addition operations are performed for each synapse during the whole inference, which significantly improves the SNN energy efficiency. We demonstrate the proposed TSC-SNN for VGG-16, ResNet-20 and ResNet-34 SNNs on datasets including CIFAR-10 (93.63% top-1), CIFAR-100 (70.97% top-1) and ImageNet (73.46% top-1 accuracy). It surpasses the best inference accuracy of the converted rate-encoded SNN with 7-14.5 times lesser inference latency, and 30-60 times fewer addition operations and memory accesses per inference across datasets.
论文检索
输入标题、作者或关键词，从 1,767 篇学术成果中精准定位
There has been an explosive demand for bringing machine learning (ML) powered intelligence into numerous Internet-of-Things (IoT) devices. However, the effectiveness of such intelligent functionality requires in-situ continuous model adaptation for adapting to new data and environments, while the on-device computing and energy resources are usually extremely constrained. Neither traditional hand-crafted (e.g., SGD, Adagrad, and Adam) nor existing meta optimizers are specifically designed to meet those challenges, as the former requires tedious hyper-parameter tuning while the latter are often costly due to the meta algorithms’ own overhead. To this end, we propose hardware-aware learning to optimize (HALO), a practical meta optimizer dedicated to resource-efficient on-device adaptation. Our HALO optimizer features the following highlights: (1) faster adaptation speed (i.e., taking fewer data or iterations to reach a specified accuracy) by introducing a new regularizer to promote empirical generalization; and (2) lower per-iteration complexity, thanks to a stochastic structural sparsity regularizer being enforced. Furthermore, the optimizer itself is designed as a very light-weight RNN and thus incurs negligible overhead. Ablation studies and experiments on five datasets, six optimizees, and two state-of-the-art (SOTA) edge AI devices validate that, while always achieving a better accuracy (↑0.46% - ↑20.28%), HALO can greatly trim down the energy cost (up to ↓60%) in adaptation, quantified using an IoT device or SOTA simulator. Codes and pre-trained models are at https://github.com/RICE-EIC/HALO .
4-bit and lower precision mobile models are required due to the ever-increasing demand for better energy efficiency in mobile devices. In this work, we report that the activation instability induced by weight quantization (AIWQ) is the key obstacle to sub-4-bit quantization of mobile networks. To alleviate the AIWQ problem, we propose a novel training method called PROgressive-Freezing Iterative Training (PROFIT), which attempts to freeze layers whose weights are affected by the instability problem stronger than the other layers. We also propose a differentiable and unified quantization method (DuQ) and a negative padding idea to support asymmetric activation functions such as h-swish. We evaluate the proposed methods by quantizing MobileNet-v1, v2, and v3 on ImageNet and report that 4-bit quantization offers comparable (within 1.48 % top-1 accuracy) accuracy to full precision baseline. In the ablation study of the 3-bit quantization of MobileNet-v3, our proposed method outperforms the state-of-the-art method by a large margin, 12.86 % of top-1 accuracy. The quantized model and source code is available at https://github.com/EunhyeokPark/PROFIT.
Optimization-based image smoothing is routinely formulated as the game between a smoothing energy and an appearance preservation energy. Achieving adequate smoothing is a fundamental goal of these image smoothing algorithms. We show that partially ""erasing"" the appearance preservation facilitate adequate image smoothing. In this paper, we call this manipulation as Erasing Appearance Preservation (EAP). We conduct an user study, allowing users to indicate the ""erasing"" positions by drawing scribbles interactively, to verify the correctness and effectiveness of EAP. We observe the characteristics of human-indicated ""erasing"" positions, and then formulate a simple and effective 0-1 knapsack to automatically synthesize the ""erasing"" positions. We test our synthesized erasing positions in a majority of image smoothing methods. Experimental results and large-scale perceptual human judgments show that the EAP solution tends to encourage the pattern separation or elimination capabilities of image smoothing algorithms. We further study the performance of the EAP solution in many image decomposition problems to decompose textures, shadows, and the challenging specular reflections. We also present examinations of diversiform image manipulation applications like texture removal, retexturing, intrinsic decomposition, layer extraction, recoloring, material manipulation, etc. Due to the widespread applicability of image smoothing, the EAP is also likely to be used in more image editing applications.
Quantization plays an important role in the energy-efficient deployment of Deep Neural Networks (DNNs) on resource-limited devices. Post-training quantization is highly desirable since it does not require retraining or access to the full training dataset. The well-established uniform scheme for post-training quantization achieves satisfactory results by converting DNNs from full-precision to 8-bit fixed-point integers. However, it suffers from significant performance degradation when quantizing to lower bit-widths. In this paper, we propose a PieceWise Linear Quantization (PWLQ) scheme to enable accurate approximation for tensor values that have bell-shaped distributions with long tails. Our approach breaks the entire quantization range into non-overlapping regions for each tensor, with each region being assigned an equal number of quantization levels. Optimal breakpoints that divide the entire range are found by minimizing the quantization error. Compared to state-of-the-art post-training quantization methods, experimental results show that our proposed method achieves superior performance on image classification, semantic segmentation, and object detection with minor overhead.
We present a novel computational imaging principle that combines diffractive optics with line (1D) sensing. When light passes through a diffraction grating, it disperses as a function of wavelength. We exploit this principle to recover 2D and even 3D positions from only line images. We derive a detailed image formation model and a learning-based algorithm for 2D position estimation. We show several extensions of our system to improve the accuracy of the 2D positioning and expand the effective field of view. We demonstrate our approach in two applications: (a) fast passive imaging of sparse light sources like street lamps, headlights at night and LED-based motion capture, and (b) structured light 3D scanning with line illumination and line sensing. Line imaging has several advantages over 2D sensors: high frame rate, high dynamic range, high fill-factor with additional on-chip computation, low-cost beyond the visible spectrum, and high energy efficiency when used with line illumination. Thus, our system is able to achieve high-speed and high-accuracy 2D positioning of light sources and 3D scanning of scenes.
Empowering Relational Network by Self-Attention Augmented Conditional Random Fields for Group Activity Recognition
PDF ↗This paper presents a novel relational network for group activity recognition. The core of our network is to augment the conditional random fields (CRF), amenable to learning inter-dependency of correlated observations, with the newly devised temporal and spatial self-attention to learn the temporal evolution and spatial relational contexts of every actor in videos. Such a combination utilizes the global receptive fields of self-attention to construct a spatio-temporal graph topology to address the temporal dependency and non-local relationships of the actors. The network first uses the temporal self-attention along with the spatial self-attention, which considers multiple cliques with different scales of locality to account for the diversity of the actors' relationships in group activities, to model the pairwise energy of CRF. Afterward, to accommodate the distinct characteristics of each video, a new mean-field inference algorithm with dynamic halting is also addressed. Finally, a bidirectional universal transformer encoder (UTE), which combines both of the forward and backward temporal context information, is used to aggregate the relational contexts and scene information for group activity recognition. Simulations show that the proposed approach surpasses the state-of-the-art methods on the widespread Volleyball and Collective Activity datasets.