Learned embeddings for products are an important building block web-scale e-commerce recommendation systems. At Pinterest, we build a single set of product called ItemSage to provide relevant recommendations in all shopping use cases including user, image and search based recommendations. This approach has led significant improvements engagement conversion metrics, while reducing both infrastructure maintenance cost. While most prior work focuses on from features coming modality, introduce...

Two continuum approximation (CA) optimization models are formulated to design city-wide transit systems at minimum cost.Transit routes assumed lie atop a city's street network.Model 1 assumes that the city streets laid out in ring-radial fashion.Model 2 form grid.Both can furnish hybrid designs, which exhibit intersecting central (downtown) district and only radial branching periphery.Model allows service frequency route spacing location vary arbitrarily with location's distance from...

Online decision-making problem requires us to make a sequence of decisions based on incremental information. Common solutions often need learn reward model different actions given the contextual information and then maximize long-term reward. It is meaningful know if posited reasonable how performs in asymptotic sense. We study this under setup bandit framework with linear model. The ε-greedy policy adopted address classic exploration-and-exploitation dilemma. Using martingale central limit...

This research investigated the value co-creation behaviors in livestreaming platforms and internal mechanism of perceived on consumer behavior short-video platform TikTok. selected Tiktok as object, uses structural equation model to analyze data. The results indicated that consumer-perceived mediates relationship between engagement citizenship behaviors. In addition, exhibit significant community attributes, interactive forms primary part enhances value. Consumers join look for communities...

Online decision making aims to learn the optimal rule by personalized decisions and updating recursively. It has become easier than before with help of big data, but new challenges also come along. Since should be updated once per step, an offline update which uses all historical data is inefficient in computation storage. To this end, we propose a completely online algorithm that can make via stochastic gradient descent. not only efficient supports kinds parametric reward models. Focusing...

In the context of urban public spaces, which tend to be shrunk and humanistic, micro spaces (MUSs) might offer residents additional opportunities for activities occur because their small scale, discrete distribution, frequent interaction. Cold cities have severe climatic restrictions; thus, as seasons change, MUSs in cold regions are more significant than medium- large-scale spaces. To investigate how contribute enhancing vitality during transition seasons, a system indicators was developed...

Cloud Storage provides external data storage service by combing and coordinating different types of devices in the network, so that they can collectively work together. However it always exists a trust game relationship between users providers, therefore building healthy, fair secure cloud environment is necessary, especially for security auditing on state operation processes. This paper proposes user behavior (UB)-based scheme, analyzing log from servers. Firstly, we present description...

Service function chaining (SFC) is able to provide customizable network services the traffic flows of different IoT subjects. Nowadays, SFC becomes profound implement service requirements devices with flexibility and programmability provided by emerging technologies, software defined (SDN) virtualization. These techniques play an increasingly important role for deployment allow requirement certain device be specified subjects, including SDN applications managers. However, independent...

Machine learning has become more important in real-life decision-making but people are concerned about the ethical problems it may bring when used improperly. Recent work brings discussion of machine fairness into causal framework and elaborates on concept Counterfactual Fairness. In this article, we develop Fair Learning through dAta Preprocessing (FLAP) algorithm to learn counterfactually fair decisions from biased training data formalize conditions where different preprocessing procedures...

Engagement analysis finds various applications in healthcare, education, advertisement, services. Deep Neural Networks, used for analysis, possess complex architecture and need large amounts of input data, computational power, inference time. These constraints challenge embedding systems into devices real-time use. To address these limitations, we present a novel two-stream feature fusion "Tensor-Convolution Convolution-Transformer Network" (TCCT-Net) architecture. better learn the...

Person re-identification (ReID) has become a critical component in various security and surveillance systems, necessitating accurate robust identification of individuals across different camera views. The motivation for enhancing ReID systems stems from the need to improve public safety, prevent crime, support forensic investigations. Despite significant advances, faces several key challenges that hinder its deployment real-world applications. This review paper addresses person (ReID):...

Persona, a special means of image management by social media influencers (SMIs), has become trending phenomenon and it is expected to substantially affect SMIs’ persuasiveness ad effect. This study aims explore the impact personas constructed through personal values on effectiveness video ads. Adopting Stimulus-Organism-Response (S-O-R) framework, this validates how self-enhancement self-transcendence consumer cognitive processes responses. Consumer behavioral intentions are used verify ads...

User-to-item retrieval has been an active research area in recommendation system, and two tower models are widely adopted due to model simplicity serving efficiency. In this work, we focus on a variant called conditional retrieval, where expect retrieved items be relevant condition (e.g. topic). We propose method that uses the same training data as standard but incorporates item-side information conditions query. This allows us bootstrap new use cases encourages feature interactions between...

Access Control System is evolved from the traditional lock, as name suggests, it used to regulate entrance channel. Password access control system, contactless card system and fingerprint most system. But these systems have many security risks: password easy leak, cards are very lose fingerprints can be easily extracted by criminals. Therefore, in this article I designed an based on dynamic audio token android operating Experiments show that not only safe, but also use, users complete...

Recently, the integrity management of large size atmospheric storage tank in China is still preliminary stage. The purpose and concept management, core technical system are discussed. Main framework about developed, combining with characters including super-large inner volume, corrosive nature media, dangerous leakage explosion, strict request for non-interruption operating, related national codes management. Two elements system, an process tanks which combines quality control, change,...

The smart mobile terminals and broadband communications become more popular. Access control systems using traditional mechanical lock, IC card authentication password technology exposes security issues. This paper describes the design implementation of server module. system used JAVA language development JDBC to connect MySQL database. also NIO communication architecture support high concurrent throughput non blocking MINA achieve network transmission data. Experiments show that is stable...

To address the problem of incomplete graph structure features due to lack contextual information in existing neural network-based vulnerability mining methods and over-smoothing that prevents model from learning higher-order resulting poor prediction performance, a detection method based on deep convolutional networks attention mechanism, PSG-GCNIIAT, is proposed. PSG fuses order relational graphs basis program dependency so code statements have ability sense their information, generates...

Machine learning has become more important in real-life decision-making but people are concerned about the ethical problems it may bring when used improperly. Recent work brings discussion of machine fairness into causal framework and elaborates on concept Counterfactual Fairness. In this paper, we develop Fair Learning through dAta Preprocessing (FLAP) algorithm to learn counterfactually fair decisions from biased training data formalize conditions where different preprocessing procedures...

Learned embeddings for products are an important building block web-scale e-commerce recommendation systems. At Pinterest, we build a single set of product called ItemSage to provide relevant recommendations in all shopping use cases including user, image and search based recommendations. This approach has led significant improvements engagement conversion metrics, while reducing both infrastructure maintenance cost. While most prior work focuses on from features coming modality, introduce...