Harbin Institute of Technology

About: Harbin Institute of Technology is a education organization based out in Harbin, China. It is known for research contribution in the topics: Microstructure & Control theory. The organization has 88259 authors who have published 109297 publications receiving 1603393 citations. The organization is also known as: HIT.

Learning Convolutional Networks for Content-weighted Image Compression

Abstract: Lossy image compression is generally formulated as a joint rate-distortion optimization to learn encoder, quantizer, and decoder. However, the quantizer is non-differentiable, and discrete entropy estimation usually is required for rate control. These make it very challenging to develop a convolutional network (CNN)-based image compression system. In this paper, motivated by that the local information content is spatially variant in an image, we suggest that the bit rate of the different parts of the image should be adapted to local content. And the content aware bit rate is allocated under the guidance of a content-weighted importance map. Thus, the sum of the importance map can serve as a continuous alternative of discrete entropy estimation to control compression rate. And binarizer is adopted to quantize the output of encoder due to the binarization scheme is also directly defined by the importance map. Furthermore, a proxy function is introduced for binary operation in backward propagation to make it differentiable. Therefore, the encoder, decoder, binarizer and importance map can be jointly optimized in an end-to-end manner by using a subset of the ImageNet database. In low bit rate image compression, experiments show that our system significantly outperforms JPEG and JPEG 2000 by structural similarity (SSIM) index, and can produce the much better visual result with sharp edges, rich textures, and fewer artifacts.

STGAN: A Unified Selective Transfer Network for Arbitrary Image Attribute Editing

Abstract: Arbitrary attribute editing generally can be tackled by incorporating encoder-decoder and generative adversarial networks. However, the bottleneck layer in encoder-decoder usually gives rise to blurry and low quality editing result. And adding skip connections improves image quality at the cost of weakened attribute manipulation ability. Moreover, existing methods exploit target attribute vector to guide the flexible translation to desired target domain. In this work, we suggest to address these issues from selective transfer perspective. Considering that specific editing task is certainly only related to the changed attributes instead of all target attributes, our model selectively takes the difference between target and source attribute vectors as input. Furthermore, selective transfer units are incorporated with encoder-decoder to adaptively select and modify encoder feature for enhanced attribute editing. Experiments show that our method (i.e., STGAN) simultaneously improves attribute manipulation accuracy as well as perception quality, and performs favorably against state-of-the-arts in arbitrary face attribute editing and season translation.

Model Approximation for Discrete-Time State-Delay Systems in the T–S Fuzzy Framework

Abstract: This paper is concerned with the problem of H∞ model approximation for discrete-time Takagi-Sugeno (T-S) fuzzy time-delay systems. For a given stable T- S fuzzy system, our attention is focused on the construction of a reduced-order model, which not only approximates the original system well in an H∞ performance but is also translated into a linear lower dimensional system. By applying the delay partitioning approach, a delay-dependent sufficient condition is proposed for the asymptotic stability with an H∞ error performance for the error system. Then, the H∞ model approximation problem is solved by using the projection approach, which casts the model approximation into a sequential minimization problem subject to linear matrix inequality (LMI) constraints by employing the cone complementary linearization algorithm. Moreover, by further extending the results, H∞ model approximation with special structures is obtained, i.e., delay-free model and zero-order model. Finally, two numerical examples are provided to illustrate the effectiveness of the proposed methods.

An Improved Distributed Secondary Control Method for DC Microgrids With Enhanced Dynamic Current Sharing Performance

Abstract: This paper proposes an improved distributed secondary control scheme for dc microgrids (MGs), aiming at overcoming the drawbacks of conventional droop control method. The proposed secondary control scheme can remove the dc voltage deviation and improve the current sharing accuracy by using voltage-shifting and slope-adjusting approaches simultaneously. Meanwhile, the average value of droop coefficients is calculated, and then it is controlled by an additional controller included in the distributed secondary control layer to ensure that each droop coefficient converges at a reasonable value. Hence, by adjusting the droop coefficient, each participating converter has equal output impedance, and the accurate proportional load current sharing can be achieved with different line resistances. Furthermore, the current sharing performance in steady and transient states can be enhanced by using the proposed method. The effectiveness of the proposed method is verified by detailed experimental tests based on a 3 × 1 kW prototype with three interface converters.