Design intelligence is an important branch of artificial intelligence (AI), focusing on the intelligent models and algorithms in creativity and design. In the context of AI 2.0, studies on design intelligence have developed rapidly. We summarize mainly the current emerging framework of design intelligence and review the state-of-the-art techniques of related topics, including user needs analysis, ideation, content generation, and design evaluation. Specifically, the models and methods of intelligence-generated content are reviewed in detail. Finally, we discuss some open problems and challenges for future research in design intelligence.
Development of mesoscale robots is gaining interest in security and surveillance domains due to their stealth and portable nature in achieving tasks. Their design and development require a host of hardware, controls, and behavioral innovations to yield fast, energy-efficient, distributed, adaptive, robust, and scalable systems. We extensively describe one such design and development process by: (1) the genealogy of our embedded platforms; (2) the key system architecture and functional layout; (3) the developed and implemented design principles for mesoscale robotic systems; (4) the various key algorithms developed for effective collective operations of mesoscale robotic swarms, with applications to urban sensing and mapping. This study includes our perception of the embedded hardware requirements for reliable operations of mesoscale robotic swarms and our description of the key innovations made in magnetic sensing, indoor localization, central pattern generator control, and distributed autonomy. Although some elements of the design process of such a complex robotic system are inevitably ad-hoc, we focus on the system-of-systems design process and the component design integration. This system-of-systems process provides a basis for developing future systems in the field, and the designs represent the state-of-the-art development that may be benchmarked against and adapted to other applications.
Accurate recognition of modern and traditional porcelain styles is a challenging issue in Cantonese porcelain management due to the large variety and complex elements and patterns. We propose a hybrid system with porcelain style identification and image recreation modules. In the identification module, prediction of an unknown porcelain sample is obtained by logistic regression of ensembled neural networks of top-ranked design signatures, which are obtained by discriminative analysis and transformed features in principal components. The synthesis module is developed based on a conditional generative adversarial network, which enables users to provide a designed mask with porcelain elements to generate synthesized images of Cantonese porcelain. Experimental results of 603 Cantonese porcelain images demonstrate that the proposed model outperforms other methods relative to precision, recall, area under curve of receiver operating characteristic, and confusion matrix. Case studies on image creation indicate that the proposed system has the potential to engage the community in understanding Cantonese porcelain and promote this intangible cultural heritage.
Artificial intelligence (AI) has played a significant role in imitating and producing large-scale designs such as e-commerce banners. However, it is less successful at creative and collaborative design outputs. Most humans express their ideas as rough sketches, and lack the professional skills to complete pleasing paintings. Existing AI approaches have failed to convert varied user sketches into artistically beautiful paintings while preserving their semantic concepts. To bridge this gap, we have developed SmartPaint, a co-creative drawing system based on generative adversarial networks (GANs), enabling a machine and a human being to collaborate in cartoon landscape painting. SmartPaint trains a GAN using triples of cartoon images, their corresponding semantic label maps, and edge detection maps. The machine can then simultaneously understand the cartoon style and semantics, along with the spatial relationships among the objects in the landscape images. The trained system receives a sketch as a semantic label map input, and automatically synthesizes its edge map for stable handling of varied sketches. It then outputs a creative and fine painting with the appropriate style corresponding to the human’s sketch. Experiments confirmed that the proposed SmartPaint system successfully generates high-quality cartoon paintings.
Multimedia content is an integral part of Alibaba’s business ecosystem and is in great demand. The production of multimedia content usually requires high technology and much money. With the rapid development of artificial intelligence (AI) technology in recent years, to meet the design requirements of multimedia content, many AI auxiliary tools for the production of multimedia content have emerged and become more and more widely used in Alibaba’s business ecology. Related applications include mainly auxiliary design, graphic design, video generation, and page production. In this report, a general pipeline of the AI auxiliary tools is introduced. Four representative tools applied in the Alibaba Group are presented for the applications mentioned above. The value brought by multimedia content design combined with AI technology has been well verified in business through these tools. This reflects the great role played by AI technology in promoting the production of multimedia content. The application prospects of the combination of multimedia content design and AI are also indicated.
As a promising physical layer technique, nonorthogonal multiple access (NOMA) can admit multiple users over the same space-time resource block, and thus improve the spectral efficiency and increase the number of access users. Specifically, NOMA provides a feasible solution to massive Internet of Things (IoT) in 5G and beyond-5G wireless networks over a limited radio spectrum. However, severe co-channel interference and high implementation complexity hinder its application in practical systems. To solve these problems, multiple-antenna techniques have been widely used in NOMA systems by exploiting the benefits of spatial degrees of freedom. This study provides a comprehensive review of various multiple-antenna techniques in NOMA systems, with an emphasis on spatial interference cancellation and complexity reduction. In particular, we provide a detailed investigation on multiple-antenna techniques in two-user, multiuser, massive connectivity, and heterogeneous NOMA systems. Finally, future research directions and challenges are identified.
We propose a nonvolatile resistive random access memory device by employing nanodispersion of zirconia (ZrO2) quantum dots (QDs) for the formation of an active layer. The memory devices comprising a typical sandwich structure of Ag (top)/ZrO2 (active layer)/Ti (bottom) are fabricated using a facile spin-coating method. The optimized device exhibits a high resistance state/low resistance state resistance difference (about 10 Ω), a good cycle performance (the number of cycles larger than 100), and a relatively low conversion current (about 1 μA). Atomic force microscopy and scanning electron microscope are used to observe the surface morphology and stacking state of the ZrO2 active layer. Experimental results show that the ZrO2 active layer is stacked compactly and has a low roughness (Ra=4.49 nm) due to the uniform distribution of the ZrO2 QDs. The conductive mechanism of the Ag/ZrO2/Ti device is analyzed and studied, and the conductive filaments of Ag ions and oxygen vacancies are focused on to clarify the resistive switching memory behavior. This study offers a facile approach of memristors for future electronic applications.
A four-dimensional memristive system is constructed using a novel ideal memristor with cosine memductance. Due to the special memductance nonlinearity, this memristive system has a line equilibrium set (0, 0, 0, δ) located along the coordinate of the inner state variable of the memristor, whose stability is periodically varied with a change of δ. Nonlinear and one-dimensional initial offset boosting behaviors, which are triggered by not only the initial condition of the memristor but also other two initial conditions, are numerically uncovered. Specifically, a wide variety of coexisting attractors with different positions and topological structures are revealed along the boosting route. Finally, circuit simulations are performed by Power SIMulation (PSIM) to confirm the unique dynamical features.
Recent attempts at key-aggregate searchable encryption (KASE) combine the advantages of searching encrypted data with support for data owners to share an aggregate searchable key with a user delegating search rights to a set of data. A user, in turn, is required to submit only one single aggregate trapdoor to the cloud to perform a keyword search across the shared set of data. However, the existing KASE methods do not support searching through data that are shared by multiple owners using a single aggregate trapdoor. Therefore, we propose a MULKASE method that allows a user to search across different data records owned by multiple users using a single trapdoor. In MULKASE, the size of the aggregate key is independent of the number of documents held by a data owner. The size of an aggregate key remains constant even though the number of outsourced ciphertexts goes beyond the predefined limit. Security analysis proves that MULKASE is secure against chosen message attacks and chosen keyword attacks. In addition, the security analysis confirms that MULKASE is secure against cross-pairing attacks and provides query privacy. Theoretical and empirical analyses show that MULKASE performs better than the existing KASE methods. We also illustrate how MULKASE can carry out federated searches.