Yong Shi

Md Romyull Islam, Bobin Deng, Nobel Dhar et al.

Cloud-based large language models (LLMs) and their variants have significantly influenced real-world applications. Deploying smaller models (i.e., small language models (SLMs)) on edge devices offers additional advantages, such as reduced latency and independence from network connectivity. However, edge devices' limited computing resources and constrained energy budgets challenge efficient deployment. This study evaluates the power efficiency of five representative SLMs - Llama 3.2, Phi-3 Mini, TinyLlama, and Gemma 2 on Raspberry Pi 5, Jetson Nano, and Jetson Orin Nano (CPU and GPU configurations). Results show that Jetson Orin Nano with GPU acceleration achieves the highest energy-to-performance ratio, significantly outperforming CPU-based setups. Llama 3.2 provides the best balance of accuracy and power efficiency, while TinyLlama is well-suited for low-power environments at the cost of reduced accuracy. In contrast, Phi-3 Mini consumes the most energy despite its high accuracy. In addition, GPU acceleration, memory bandwidth, and model architecture are key in optimizing inference energy efficiency. Our empirical analysis offers practical insights for AI, smart systems, and mobile ad-hoc platforms to leverage tradeoffs from accuracy, inference latency, and power efficiency in energy-constrained environments.

Jiawen Xiong, Yong Shi, Boyuan Chen et al.

Build verifiability refers to the property that the build of a software system can be verified by independent third parties and it is crucial for the trustworthiness of a software system. Various efforts towards build verifiability have been made to C/C++-based systems, yet the techniques for Java-based systems are not systematic and are often specific to a particular build tool (e.g., Maven). In this study, we present a systematic approach towards build verifiability on Java-based systems. Our approach consists of three parts: a unified build process, a tool that dynamically controls non-determinism during the build process, and another tool that eliminates non-equivalences by post-processing the build artifacts. We apply our approach on 46 unverified open source projects from Reproducible Central and 13 open source projects that are widely used by Huawei commercial products. As a result, 91% of the unverified Reproducible Central projects and 100% of the commercially adopted OSS projects are successfully verified with our approach. In addition, based on our experience in analyzing thousands of builds for both commercial and open source Java-based systems, we present 14 patterns that introduce non-equivalences in generated build artifacts and their respective mitigation strategies. Among these patterns, 11 (78%) are unique for Java-based system, whereas the remaining 3 (22%) are common with C/C++-based systems. The approach and the findings of this paper are useful for both practitioners and researchers who are interested in build verifiability.

Boyuan Chen, Mingzhi Wen, Yong Shi et al.

Reproducibility is an increasing concern in Artificial Intelligence (AI), particularly in the area of Deep Learning (DL). Being able to reproduce DL models is crucial for AI-based systems, as it is closely tied to various tasks like training, testing, debugging, and auditing. However, DL models are challenging to be reproduced due to issues like randomness in the software (e.g., DL algorithms) and non-determinism in the hardware (e.g., GPU). There are various practices to mitigate some of the aforementioned issues. However, many of them are either too intrusive or can only work for a specific usage context. In this paper, we propose a systematic approach to training reproducible DL models. Our approach includes three main parts: (1) a set of general criteria to thoroughly evaluate the reproducibility of DL models for two different domains, (2) a unified framework which leverages a record-and-replay technique to mitigate software-related randomness and a profile-and-patch technique to control hardware-related non-determinism, and (3) a reproducibility guideline which explains the rationales and the mitigation strategies on conducting a reproducible training process for DL models. Case study results show our approach can successfully reproduce six open source and one commercial DL models.

Minglong Lei, Yong Shi, Lingfeng Niu

Graph auto-encoders have proved to be useful in network embedding task. However, current models only consider explicit structures and fail to explore the informative latent structures cohered in networks. To address this issue, we propose a latent network embedding model based on adversarial graph auto-encoders. Under this framework, the problem of discovering latent structures is formulated as inferring the latent ties from partial observations. A latent transmission matrix that describes the strengths of existing edges and latent ties is derived based on influence cascades sampled by simulating diffusion processes over networks. Besides, since the inference process may bring extra noises, we introduce an adversarial training that works as regularization to dislodge noises and improve the model robustness. Extensive experiments on link prediction and node classification tasks show that the proposed model achieves superior results compared with baseline models.

Jie Yang, Ruijie Xu, Zhiquan Qi et al.

Visual anomaly detection is an important and challenging problem in the field of machine learning and computer vision. This problem has attracted a considerable amount of attention in relevant research communities. Especially in recent years, the development of deep learning has sparked an increasing interest in the visual anomaly detection problem and brought a great variety of novel methods. In this paper, we provide a comprehensive survey of the classical and deep learning-based approaches for visual anomaly detection in the literature. We group the relevant approaches in view of their underlying principles and discuss their assumptions, advantages, and disadvantages carefully. We aim to help the researchers to understand the common principles of visual anomaly detection approaches and identify promising research directions in this field.

Jie Yang, Yong Shi, Zhiquan Qi

Automatic detecting anomalous regions in images of objects or textures without priors of the anomalies is challenging, especially when the anomalies appear in very small areas of the images, making difficult-to-detect visual variations, such as defects on manufacturing products. This paper proposes an effective unsupervised anomaly segmentation approach that can detect and segment out the anomalies in small and confined regions of images. Concretely, we develop a multi-scale regional feature generator that can generate multiple spatial context-aware representations from pre-trained deep convolutional networks for every subregion of an image. The regional representations not only describe the local characteristics of corresponding regions but also encode their multiple spatial context information, making them discriminative and very beneficial for anomaly detection. Leveraging these descriptive regional features, we then design a deep yet efficient convolutional autoencoder and detect anomalous regions within images via fast feature reconstruction. Our method is simple yet effective and efficient. It advances the state-of-the-art performances on several benchmark datasets and shows great potential for real applications.

Jie Yang, Zhiquan Qi, Yong Shi

This paper develops a multi-task learning framework that attempts to incorporate the image structure knowledge to assist image inpainting, which is not well explored in previous works. The primary idea is to train a shared generator to simultaneously complete the corrupted image and corresponding structures --- edge and gradient, thus implicitly encouraging the generator to exploit relevant structure knowledge while inpainting. In the meantime, we also introduce a structure embedding scheme to explicitly embed the learned structure features into the inpainting process, thus to provide possible preconditions for image completion. Specifically, a novel pyramid structure loss is proposed to supervise structure learning and embedding. Moreover, an attention mechanism is developed to further exploit the recurrent structures and patterns in the image to refine the generated structures and contents. Through multi-task learning, structure embedding besides with attention, our framework takes advantage of the structure knowledge and outperforms several state-of-the-art methods on benchmark datasets quantitatively and qualitatively.

Biao Li, Jiabin Liu, Bo Wang et al.

Deep learning (DL) architectures for superresolution (SR) normally contain tremendous parameters, which has been regarded as the crucial advantage for obtaining satisfying performance. However, with the widespread use of mobile phones for taking and retouching photos, this character greatly hampers the deployment of DL-SR models on the mobile devices. To address this problem, in this paper, we propose a super lightweight SR network: s-LWSR. There are mainly three contributions in our work. Firstly, in order to efficiently abstract features from the low resolution image, we build an information pool to mix multi-level information from the first half part of the pipeline. Accordingly, the information pool feeds the second half part with the combination of hierarchical features from the previous layers. Secondly, we employ a compression module to further decrease the size of parameters. Intensive analysis confirms its capacity of trade-off between model complexity and accuracy. Thirdly, by revealing the specific role of activation in deep models, we remove several activation layers in our SR model to retain more information for performance improvement. Extensive experiments show that our s-LWSR, with limited parameters and operations, can achieve similar performance to other cumbersome DL-SR methods.

Jiabin Liu, Bo Wang, Zhiquan Qi et al.

In this paper, we leverage generative adversarial networks (GANs) to derive an effective algorithm LLP-GAN for learning from label proportions (LLP), where only the bag-level proportional information in labels is available. Endowed with end-to-end structure, LLP-GAN performs approximation in the light of an adversarial learning mechanism, without imposing restricted assumptions on distribution. Accordingly, we can directly induce the final instance-level classifier upon the discriminator. Under mild assumptions, we give the explicit generative representation and prove the global optimality for LLP-GAN. Additionally, compared with existing methods, our work empowers LLP solver with capable scalability inheriting from deep models. Several experiments on benchmark datasets demonstrate vivid advantages of the proposed approach.

Yong Shi, Huadong Wang, Xin Shen et al.

Ordinal regression (OR) is a special multiclass classification problem where an order relation exists among the labels. Recent years, people share their opinions and sentimental judgments conveniently with social networks and E-Commerce so that plentiful large-scale OR problems arise. However, few studies have focused on this kind of problems. Nonparallel Support Vector Ordinal Regression (NPSVOR) is a SVM-based OR model, which learns a hyperplane for each rank by solving a series of independent sub-optimization problems and then ensembles those learned hyperplanes to predict. The previous studies are focused on its nonlinear case and got a competitive testing performance, but its training is time consuming, particularly for large-scale data. In this paper, we consider NPSVOR's linear case and design an efficient training method based on the dual coordinate descent method (DCD). To utilize the order information among labels in prediction, a new prediction function is also proposed. Extensive contrast experiments on the text OR datasets indicate that the carefully implemented DCD is very suitable for training large data.

Zilong Lin, Yong Shi, Zhi Xue

As an essential tool in security, the intrusion detection system bears the responsibility of the defense to network attacks performed by malicious traffic. Nowadays, with the help of machine learning algorithms, intrusion detection systems develop rapidly. However, the robustness of this system is questionable when it faces adversarial attacks. For the robustness of detection systems, more potential attack approaches are under research. In this paper, a framework of the generative adversarial networks, called IDSGAN, is proposed to generate the adversarial malicious traffic records aiming to attack intrusion detection systems by deceiving and evading the detection. Given that the internal structure and parameters of the detection system are unknown to attackers, the adversarial attack examples perform the black-box attacks against the detection system. IDSGAN leverages a generator to transform original malicious traffic records into adversarial malicious ones. A discriminator classifies traffic examples and dynamically learns the real-time black-box detection system. More significantly, the restricted modification mechanism is designed for the adversarial generation to preserve original attack functionalities of adversarial traffic records. The effectiveness of the model is indicated by attacking multiple algorithm-based detection models with different attack categories. The robustness is verified by changing the number of the modified features. A comparative experiment with adversarial attack baselines demonstrates the superiority of our model.

Feng Liu, Yong Shi

The unclear development direction of human society is a deep reason for that it is difficult to form a uniform ethical standard for human society and artificial intelligence. Since the 21st century, the latest advances in the Internet, brain science and artificial intelligence have brought new inspiration to the research on the development direction of human society. Through the study of the Internet brain model, AI IQ evaluation, and the evolution of the brain, this paper proposes that the evolution of population knowledge base is the key for judging the development direction of human society, thereby discussing the standards and norms for the construction of artificial intelligence ethics.

Yong Shi, Minglong Lei, Peng Zhang et al.

In network embedding, random walks play a fundamental role in preserving network structures. However, random walk based embedding methods have two limitations. First, random walk methods are fragile when the sampling frequency or the number of node sequences changes. Second, in disequilibrium networks such as highly biases networks, random walk methods often perform poorly due to the lack of global network information. In order to solve the limitations, we propose in this paper a network diffusion based embedding method. To solve the first limitation, our method employs a diffusion driven process to capture both depth information and breadth information. The time dimension is also attached to node sequences that can strengthen information preserving. To solve the second limitation, our method uses the network inference technique based on cascades to capture the global network information. To verify the performance, we conduct experiments on node classification tasks using the learned representations. Results show that compared with random walk based methods, diffusion based models are more robust when samplings under each node is rare. We also conduct experiments on a highly imbalanced network. Results shows that the proposed model are more robust under the biased network structure.

Yunlong Mi, Yong Shi, Jinhai Li

Concept-cognitive learning (CCL) is a hot topic in recent years, and it has attracted much attention from the communities of formal concept analysis, granular computing and cognitive computing. However, the relationship among cognitive computing (CC), concept-cognitive computing (CCC), CCL and concept-cognitive learning model (CCLM) is not clearly described. To this end, we first explain the relationship of CC, CCC, CCL and CCLM. Then, we propose a generalized concept-cognitive learning (GCCL) from the point of view of machine learning. Finally, experiments on some data sets are conducted to verify the feasibility of concept formation and concept-cognitive process of GCCL.

Feng Liu, Yong Shi, Ying Liu

The rapid development of artificial intelligence has brought the artificial intelligence threat theory as well as the problem about how to evaluate the intelligence level of intelligent products. Both need to find a quantitative method to evaluate the intelligence level of intelligence systems, including human intelligence. Based on the standard intelligence system and the extended Von Neumann architecture, this paper proposes General IQ, Service IQ and Value IQ evaluation methods for intelligence systems, depending on different evaluation purposes. Among them, the General IQ of intelligence systems is to answer the question of whether the artificial intelligence can surpass the human intelligence, which is reflected in putting the intelligence systems on an equal status and conducting the unified evaluation. The Service IQ and Value IQ of intelligence systems are used to answer the question of how the intelligent products can better serve the human, reflecting the intelligence and required cost of each intelligence system as a product in the process of serving human.

Feng Liu, Yong Shi, Ying Liu

Although artificial intelligence is currently one of the most interesting areas in scientific research, the potential threats posed by emerging AI systems remain a source of persistent controversy. To address the issue of AI threat, this study proposes a standard intelligence model that unifies AI and human characteristics in terms of four aspects of knowledge, i.e., input, output, mastery, and creation. Using this model, we observe three challenges, namely, expanding of the von Neumann architecture; testing and ranking the intelligence quotient of naturally and artificially intelligent systems, including humans, Google, Bing, Baidu, and Siri; and finally, the dividing of artificially intelligent systems into seven grades from robots to Google Brain. Based on this, we conclude that AlphaGo belongs to the third grade.

Feng Liu, Yong Shi

Currently, potential threats of artificial intelligence (AI) to human have triggered a large controversy in society, behind which, the nature of the issue is whether the artificial intelligence (AI) system can be evaluated quantitatively. This article analyzes and evaluates the challenges that the AI development level is facing, and proposes that the evaluation methods for the human intelligence test and the AI system are not uniform; and the key reason for which is that none of the models can uniformly describe the AI system and the beings like human. Aiming at this problem, a standard intelligent system model is established in this study to describe the AI system and the beings like human uniformly. Based on the model, the article makes an abstract mathematical description, and builds the standard intelligent machine mathematical model; expands the Von Neumann architecture and proposes the Liufeng - Shiyong architecture; gives the definition of the artificial intelligence IQ, and establishes the artificial intelligence scale and the evaluation method; conduct the test on 50 search engines and three human subjects at different ages across the world, and finally obtains the ranking of the absolute IQ and deviation IQ ranking for artificial intelligence IQ 2014.

Feng Liu, Yong Shi

Whether the machine intelligence can surpass the human intelligence is a controversial issue. On the basis of traditional IQ, this article presents the Universal IQ test method suitable for both the machine intelligence and the human intelligence. With the method, machine and human intelligences were divided into 4 major categories and 15 subcategories. A total of 50 search engines across the world and 150 persons at different ages were subject to the relevant test. And then, the Universal IQ ranking list of 2014 for the test objects was obtained.