Malicious URL, a.k.a. malicious website, is a common and serious threat to cybersecurity. URLs host unsolicited content (spam, phishing, drive-by exploits, etc.) lure unsuspecting users become victims of scams (monetary loss, theft private information, malware installation), cause losses billions dollars every year. It imperative detect act on such threats in timely manner. Traditionally, this detection done mostly through the usage blacklists. However, blacklists cannot be exhaustive, lack...

Autoregressive integrated moving average (ARIMA) is one of the most popular linear models for time series forecasting due to its nice statistical properties and great flexibility. However, parameters are estimated in a batch manner noise terms often assumed be strictly bounded, which restricts applications makes it inefficient handling large-scale real data. In this paper, we propose online learning algorithms estimating ARIMA under relaxed assumptions on terms, suitable wider range enjoys...

Food computing is playing an increasingly important role in human daily life, and has found tremendous applications guiding behavior towards smart food consumption healthy lifestyle. An task under the food-computing umbrella retrieval, which particularly helpful for health related applications, where we are interested retrieving information about (e.g., ingredients, nutrition, etc.). In this paper, investigate open research of cross-modal retrieval between cooking recipes images, propose a...

Low bit-rate speech codecs have been widely used in audio communications like VoIP and mobile communications, so that steganography low streams would broad applications practice. In this paper, the authors propose a new algorithm for by integrating information hiding into process of encoding. The proposed performs data embedding while pitch period prediction is conducted during encoding, thus maintaining synchronization between can achieve high quality prevent detection steganalysis, but...

Attributed networks nowadays are ubiquitous in a myriad of high-impact applications, such as social network analysis, financial fraud detection, and drug discovery. As central analytical task on attributed networks, node classification has received much attention the research community. In real-world large portion classes only contains limited labeled instances, rendering long-tail class distribution. Existing algorithms unequipped to handle few-shot classes. remedy, learning attracted surge...

Node classification is an important problem on graphs. While recent advances in graph neural networks achieve promising performance, they require abundant labeled nodes for training. However, many practical scenarios, there often exist novel classes which only one or a few are available as supervision, known few-shot node classification. Although meta-learning has been widely used vision and language domains to address learning, its adoption graphs limited. In particular, task not...

Recent studies have highlighted the limitations of message-passing based graph neural networks (GNNs), e.g., limited model expressiveness, over-smoothing, over-squashing, etc. To alleviate these issues, Graph Transformers (GTs) been proposed which work in paradigm that allows message passing to a larger coverage even across whole graph. Hinging on global range attention mechanism, GTs shown superpower for representation learning homogeneous graphs. However, investigation heterogeneous...

Attributed networks are pervasive in numerous of high-impact domains. As opposed to conventional plain where only pairwise node dependencies observed, both the network topology and attribute information readily available on attributed networks. More often than not, nodal attributes depicted a high-dimensional feature space therefore notoriously difficult tackle due curse dimensionality. Additionally, features that irrelevant structure could hinder discovery actionable patterns from Hence, it...

Nowadays, driven by the increasing concern on diet and health, food computing has attracted enormous attention from both industry research community. One of most popular topics in this domain is Food Retrieval, due to its profound influence health-oriented applications. In paper, we focus task cross-modal retrieval between images cooking recipes. We present Modality-Consistent Embedding Network (MCEN) that learns modality-invariant representations projecting texts same embedding space. To...

As important side information, attributes have been widely exploited in the existing recommender system for better performance. However, real-world scenarios, it is common that some of items/users are missing (e.g., movies miss genre data). Prior studies usually use a default value (i.e., “other”) to represent attribute, resulting sub-optimal To address this problem, paper, we present an attribute-aware attentive graph convolution network (A <inline-formula><tex-math...

Synthetic lethality (SL) is a promising concept for novel discovery of anti-cancer drug targets. However, wet-lab experiments detecting SLs are faced with various challenges, such as high cost, low consistency across platforms, or cell lines. Therefore, computational prediction methods needed to address these issues. This paper proposes SL method, named <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> MF, which employs logistic matrix...

Food retrieval is an important task to perform analysis of food-related information, where we are interested in retrieving relevant information about the queried food item such as ingredients, cooking instructions, etc. In this paper, investigate cross-modal between images and recipes. The goal learn embedding recipes a common feature space, that corresponding image-recipe embeddings lie close one another. Two major challenges addressing problem 1) large intra-variance small inter-variance...

Tumor necrosis factor α (TNF-α) is upregulated in a chronic inflammatory environment, including tumors, and has been recognized as pro-tumor many cancers. Applying the traditional TNF-α antibodies that neutralize activity, however, only exerts modest anti-tumor efficacy clinical studies. Here, we develop an innovative approach to target distinct from neutralization mechanism. We employed phage display yeast select non-neutralizing can piggyback on co-internalize into cells through receptor...

In order to develop an underwater sea cucumber collecting robot, it is necessary use the machine vision method realize recognition and location. An identification location of based on improved You Only Look Once version 5 (YOLOv5) proposed. Due low contrast between cucumbers environment, Multi-Scale Retinex with Color Restoration (MSRCR) algorithm was introduced process images enhance contrast. improve precision efficiency, Convolutional Block Attention Module (CBAM) added. make small target...

Artificial Intelligence for IT operations (AIOps) aims to combine the power of AI with big data generated by Operations processes, particularly in cloud infrastructures, provide actionable insights primary goal maximizing availability. There are a wide variety problems address, and multiple use-cases, where capabilities can be leveraged enhance operational efficiency. Here we review AIOps vision, trends challenges opportunities, specifically focusing on underlying techniques. We discuss...

Social recommendation has been playing an important role in suggesting items to users through utilizing information from social connections. However, most existing approaches do not consider the attention factor causing constraint that people can only accept a limited amount of due strength mind, which discovered as intrinsic physiological property human by science. We address this issue resorting concept science and combining it with machine learning techniques elegant way. When introducing...

Existing continual learning methods use Batch Normalization (BN) to facilitate training and improve generalization across tasks. However, the non-i.i.d non-stationary nature of data, especially in online setting, amplify discrepancy between testing BN hinder performance older In this work, we study cross-task normalization effect where normalizes data using moments biased towards current task, resulting higher catastrophic forgetting. This limitation motivates us propose a simple yet...

Federated Learning (FL) has been widely concerned for it enables decentralized learning while ensuring data privacy. However, most existing methods unrealistically assume that the classes encountered by local clients are fixed over time. After new classes, this impractical assumption will make model's catastrophic forgetting of old significantly severe. Moreover, due to limitation communication cost, is challenging use large-scale models in FL, which affect prediction accuracy. To address...

According to the Complementary Learning Systems (CLS) theory (McClelland et al. 1995) in neuroscience, humans do effective continual learning through two complementary systems: a fast system centered on hippocampus for rapid of specifics, individual experiences; and slow located neocortex gradual acquisition structured knowledge about environment. Motivated by this theory, we propose DualNets (for Dual Networks), general framework comprising supervised pattern-separated representation from...