Search

There are two general approaches to developing artificial general intelligence (AGI)1: computer-science-oriented and neuroscience-oriented. Because of the fundamental differences in their formulations and coding schemes, these two approaches rely on distinct and incompatible platforms2–8, retarding the development of AGI. A general platform that could support the prevailing computer-science-based artificial neural networks as well as neuroscience-inspired models and algorithms is highly desirable. Here we present the Tianjic chip, which integrates the two approaches to provide a hybrid, synergistic platform. The Tianjic chip adopts a many-core architecture, reconfigurable building blocks and a streamlined dataflow with hybrid coding schemes, and can not only accommodate computer-science-based machine-learning algorithms, but also easily implement brain-inspired circuits and several coding schemes. Using just one chip, we demonstrate the simultaneous processing of versatile algorithms and models in an unmanned bicycle system, realizing real-time object detection, tracking, voice control, obstacle avoidance and balance control. Our study is expected to stimulate AGI development by paving the way to more generalized hardware platforms. The ‘Tianjic’ hybrid electronic chip combines neuroscience-oriented and computer-science-oriented approaches to artificial general intelligence, demonstrated by controlling an unmanned bicycle.

Increasingly there is a need to process graphs that are larger than the available memory on today’s machines. Many systems have been developed with graph representations that are efficient and compact for out-of-core processing. A necessary task in these systems is memory management. This paper presents a system called Cacheap which automatically and efficiently manages the available memory to maximize the speed of graph processing, minimize the amount of disk access, and maximize the utilization of memory for graph data. It has a simple interface that can be easily adopted by existing graph engines. The paper describes the new system, uses it in recent graph engines, and demonstrates its integer factor improvements in the speed of large-scale graph processing.

The lensless on-chip microscope is an emerging technology in the recent decade that can realize the imaging and analysis of biological samples with a wide field-of-view without huge optical devices and any lenses. Because of its small size, low cost, and being easy to hold and operate, it can be used as an alternative tool for large microscopes in resource-poor or remote areas, which is of great significance for the diagnosis, treatment, and prevention of diseases. To improve the low-resolution characteristics of the existing lensless shadow imaging systems and to meet the high-resolution needs of point-of-care testing, here, we propose a high-precision on-chip microscope based on in-line holographic technology. We demonstrated the ability of the iterative phase recovery algorithm to recover sample information and evaluated it with image quality evaluation algorithms with or without reference. The results showed that the resolution of the holographic image after iterative phase recovery is 1.41 times that of traditional shadow imaging. Moreover, we used machine learning tools to identify and count the mixed samples of mouse ascites tumor cells and micro-particles that were iterative phase recovered. The results showed that the on-chip cell counter had high-precision counting characteristics as compared with manual counting of the microscope reference image. Therefore, the proposed high-precision lensless microscope on a chip based on in-line holographic imaging provides one promising solution for future point-of-care testing (POCT).

Evolving graphs in the real world are large-scale and constantly changing, as hundreds of thousands of updates may come every second. Monotonic algorithms such as Reachability and Shortest Path are widely used in real-time analytics to gain both static and temporal insights and can be accelerated by incremental computing. Existing streaming systems adopt the incremental computing model and achieve either low latency or high throughput, but not both. However, both high throughput and low latency are required in real scenarios such as financial fraud detection. This paper presents RisGraph, a real-time streaming system that provides low-latency analysis for each update with high throughput. RisGraph addresses the challenge with localized data access and inter-update parallelism. We propose a data structure named Indexed Adjacency Lists and use sparse arrays and Hybrid Parallel Mode to enable localized data access. To achieve inter-update parallelism, we propose a domain-specific concurrency control mechanism based on the classification of safe and unsafe updates. Experiments show that RisGraph can ingest millions of updates per second for graphs with several hundred million vertices and billions of edges, and the P999 processing time latency is within 20 milliseconds. RisGraph achieves orders-of-magnitude improvement on throughput when analyses are executed for each update without batching and performs better than existing systems with batches of up to 20 million updates.

Deep neural networks have been achieving booming breakthroughs in various artificial intelligence tasks, however they are notorious for consuming unbearable hardware resources, training time and power. The emerging pruning/binarization methods, which aim at both decreasing overheads and retaining high performance, seem to promise applications on portable devices. However, even with these most advanced algorithms, we have to save the full-precision weights during the gradient descent process which remains size and power bottlenecks of memory access and the resulting computation. To address this challenge, we propose a unified discrete state transition (DST) framework by introducing a probabilistic projection operator that constrains the weight matrices in a discrete weight space (DWS) with configurable number of states, throughout the whole training process. The experimental results over various data sets including MNIST, CIFAR10 and SVHN show the effectiveness of this framework. The direct transition between discrete states significantly saves memory for storing weights in full precision, as well as simplifies the computation of weight updating. The proposed DST framework is hardware friendly as it can be easily implemented by a wide range of emerging portable devices, including binary, ternary and multiple-level memory devices. This work paves the way for on-chip learning on various portable devices in the near future.

A novel effervescent tablet-assisted switchable hydrophilicity solvent-based on-site dispersive liquid-liquid microextraction coupled with gas chromatography detection has been developed to determine five pyrethroid insecticides in environmental water samples. In this method, effervescent tablets contain effervescent precursors and extractants, which play a dual role of dispersion and extraction. After effervescence reaction, the formation of extractant and the extraction of analytes were accomplished simultaneously. Phase separation were carried by filtration using homemade filter column, which of filter is oil-absorbing cotton polypropylene. After phase separation, the extractant phase were obtained by elution. No electric instrument was required in sample pre-treatment step in the developed method. Experimental conditions affecting the extraction efficiencies have been optimized, such as type and amount of extractant, type of effervescent tablets, etc. Under the optimized conditions, the linearity was 5-500μg L-1, with coefficients greater than 0.9990. The limit of detection and the limit of quantitation were in the range of 0.22-1.88μg L-1 and 0.75-6.25μg L-1. The relative standard deviations ranged from 0.8% to 6.1%. The enrichment factors were in the range of 65-108. This method was successfully applied to the on-site processing of tap, reservoir and river water.

Weibo is the Twitter counterpart in China that has attracted hundreds of millions of users. We crawled an almost complete Weibo user network that contains 222 million users and 27 billion links in 2013. This paper analyzes the structural properties of this network, and compares it with a Twitter user network. The topological properties we studied include the degree distributions, connected components, distance distributions, reciprocity, clustering coefficient, PageRank centrality, and degree assortativity. We find that Weibo users have a higher diversity index, higher Gini index, but a lower reciprocity and clustering coefficient for most of the nodes. A surprising observation is that the reciprocity of Weibo is only about a quarter of the reciprocity of the Twitter user network. We also show that Weibo adoption rate correlates with economic development positively, and Weibo network can be used to quantify the connections between provinces and regions in China. In particular, point-wise mutual information is shown to be accurate in quantifying the strength of connections. We developed an interactive analyzing software framework for this study, and released the data and code online.

Cadmium (Cd) pollution is a widespread threat to marine life, and ongoing ocean acidification (OA) is predicted to impact bio-toxicity of Cd compounds. However, the cascading effects of changed Cd toxicity to marine primary producers are not well characterized. Here, we studied the impact of OA on Cd toxicity responses in a globally important diatom Phaeodactylum tricornutum under both ambient and elevated pCO2 conditions. We found that increased pCO2 alleviated the impact of additive Cd toxicity on P. tricornutum not only under controlled indoor experiments but also in outdoor mesocosm experiments that reflect more natural growth conditions. Transcriptome analysis suggested that genes involved in Cd efflux and phytochelatin production were up-regulated and genes involved in Cd influx were down-regulated in long-term selected lineages under elevated pCO2. We further found a significant reduction of Cd transfer across trophic level, when the scallop Argopecten irradians was fed with Cd-exposed P. tricornutum previously cultured under elevated pCO2. Our results indicate that after long-term selection of P. tricornutum exposed to future OA conditions (i.e. elevated pCO2), the diatom alters its Cd detoxification strategy, which could have broader impacts on the bio-geochemical cycle of Cd in the marine ecosystem.

Both of the transient thermal transfer model and the thermal elastic-plastic model were adopted for the thermal and stress simulation during the additively manufacture of 2Cr13 thin-wall part. The microstructural evolution and internal defect distribution were also characterized, as well as the measurement of thermal cycles and residual stress. A fair agreement between the simulated and experimental results, including the thermal cycling curves and residual stress distribution, was achieved. During the arc deposition, the spatially transient stress distribution was evolved regularly as a consequence of the repeated transient thermal history, which finally transformed into the residual stress within the whole part and caused the wrapping phenomenon. In the middle region, the location-related metal-stable microstructure and nonhomogeneous micro-sized pore distribution was obtained within a single layer, supposedly a result of complicated in-situ micro-heating treatment when new layers continuously deposited on the top of the buildup. However, the whole top-most layer was featured by the homogenous microstructure and pore distribution, as this area didn’t undergo the further in-situ thermal cycle. The intrinsic relationship of “thermal history- microstructural evolution- defect distribution” was explored with the combination of numerical simulation and experimental characterization during the arc deposition of 2Cr13 part.

Density functional theory calculations were performed to gain insight into the mechanism and kinetic studies of homogeneous gas-phase formation of polychlorinated dibenzodioxins and polychlorinated dibenzofurans (PCDD/Fs) via aliphatic hydrocarbons (C2H2, C2H4, C3H6 and C4H8). The calculated results demonstrated that the intra-annular elimination of H is the rate-determining step throughout the reaction chain; the presence of ortho-Cl increases the abstraction barrier of arene H and decreases the reactivity of the molecule. The phenoxy radicals undergoes dimerization via carbon-carbon or carbon-oxygen coupling to form PCDD/Fs and the two coupling pathways are competitive. Our work indicates that aliphatic hydrocarbons are less reactive precursors in PCDD/F formation compared with chlorophenoxy radicals and phenoxy radicals among primary precursors of PCDD/Fs. The results presented here could be used to evaluate the contribution of aliphatic hydrocarbons acting as precursors to PCDD/Fs formation.

Inferring emotions from image social networks is a hot research topic nowadays. For image social networks (Flickr, Instagram), there is an interesting phenomenon that people would like to establish or attend virtual groups and share images with different topics and emotions in different groups. Previous researches on inferring emotions usually focus on image content and user personalization, thus leading an interesting but challenging problem: whether virtual groups can influence members(users)‘ emotions. In this paper, we systematically study this problem from two aspects: 1) whether group homophily in users’ emotions exists in image social networks; 2) how to model this subtle and complex group homophily in image social networks. Inspired by the study results of two aspects, we introduce group information to infer emotions in image social networks, and propose a novel Group-Based Factor Graph Model (G-FGM), incorporating image content, user personalization and group information to understand the emotions behind social images better. The experimental results on a dataset containing 218, 816 emotion-labeled images from Flickr show that our model outperforms (8.6-19.4% improvement in terms of F1-Measure) several baseline methods.