Reduce Search Costs Research Articles

Hessian-based mixed-precision quantization with transition aware training for neural networks.

Model quantization is widely used to realize the promise of ubiquitous embedded deep network inference. While mixed-precision quantization has shown promising performance, existing approaches often rely on time-consuming search process to determine the optimal bit configuration. To address this, we introduce Hessian-based Mixed-Precision Quantization Aware Training(HMQAT) to decrease the search overhead of bit configuration. By using the sensitivity metric of joint average Hessian trace and parameter size, HMQAT effectively guides the search process. We solve the optimization problem about bit configuration automatically using a Pareto frontier method. The lowest search overhead can be achieved through our scheme. Additionally, our approach incorporates quantization transition aware fine-tuning of scale factor. This strategy consistently ensures optimal inference performance along the accuracy-size Pareto frontier across multiple models. We extensively evaluated our method on ImageNet and CIFAR10. In particular, we show that compared to the baseline, HMQAT achieves a 10.34× reduction in model size while retaining 99.81% of the Top-1 accuracy on ResNet18 for ImageNet. Moreover, HMQAT surpassed the state-of-the-art mixed-precision quantization methods in compressing neural networks with reduced search cost and achieving a satisfying trade-off between size and accuracy. This study paves the way of deploying neural networks on lightweight devices.

Implementing Grover’s on AES-based AEAD schemes

Extensive research is currently underway to determine the security of existing ciphers in light of the advancements in quantum computing. Against symmetric key cryptography, Grover’s search algorithm is a prominent attack, capable of reducing search costs to the square root. For using Grover’s algorithm, it is imperative to embed the target cipher into a quantum circuit. Even so, this area of research is relatively new; it has garnered significant attention from the research community. In this study, we provide the first estimation of the cost of Grover’s key search attack against the AES-based AEAD schemes Rocca-S, AEGIS-128, and Tiaoxin-346. Our analysis considers circuit depth restrictions specified in NIST’s PQC standardization process. Considering NIST’s maximum depth constraints, We present the overall cost of these attacks using gate count and depth-times-width metrics. We observed that for MAXDEPTH=240\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ extsf{MAXDEPTH} = 2^{40}$$\\end{document}, Rocca-S, AEGIS-128, and Tiaoxin-346 can be retrieved using Grover’s search algorithm with gate count of 1.09 × 2253, 1.14 × 2124, and 1.22 × 2124 respectively. Concerning the current updated values by NIST, these ciphers are secure in terms of the cost of implementing Grover’s attack for key recovery. The quantum circuits of these ciphers are implemented using QISKIT, an open-source software development kit (SDK) designed for working with quantum computers running on the IBM Quantum Experience platform.

The rising entropy of English in the attention economy

We present evidence that the word entropy of American English has been rising steadily since around 1900. We also find differences in word entropy between media categories, with short-form media such as news and magazines having higher entropy than long-form media, and social media feeds having higher entropy still. To explain these results we develop an ecological model of the attention economy that combines ideas from Zipf’s law and information foraging. In this model, media consumers maximize information utility rate taking into account the costs of information search, while media producers adapt to technologies that reduce search costs, driving them to generate higher entropy content in increasingly shorter formats.

Does cross-border e-commerce improve China’s imported product quality? – A quasi-natural experiment based on the establishment of cross-border e-commerce comprehensive pilot zones

We take the establishment of cross-border e-commerce comprehensive pilot zones as the quasi-natural experiment, based on China’s General Administration of Customs Database, China Customs Online Inquiry Platform, ‘Cross-Border E-Commerce Retail Import Commodity List’ and other data. By employing the staggered DID method, this paper systematically investigates the potential of cross-border e-commerce to enhance the quality of imported products. The research findings reveal the following: Since the introduction of the cross-border e-commerce comprehensive pilot zones, there has been a significant improvement in the quality of imported products in China. These effects, however, vary across different product types and regions, suggesting heterogeneous impacts. The cross-border e-commerce comprehensive pilot zones have facilitated the development of import trade in China through three main avenues: reducing search costs, resolving financing challenges, and lowering tariffs. These measures have collectively contributed to the reduction of trade costs.

AG-SpTRSV: An Automatic Framework to Optimize Sparse Triangular Solve on GPUs

Sparse Triangular Solve (SpTRSV) has long been an essential kernel in the field of scientific computing. Due to its low computational intensity and internal data dependencies, SpTRSV is hard to implement and optimize on GPUs. Based on our experimental observations, existing implementations on GPUs fail to achieve the optimal performance due to their sub-optimal parallelism setups and code implementations, and lack of consideration of the irregular data distribution. Moreover, their algorithm design lacks the adaptability to different input matrices, which may involve substantial manual efforts of algorithm redesigning and parameter tuning for performance consistency. In this work, we propose AG-SpTRSV, an automatic framework to optimize SpTRSV on GPUs, which provides high performance on various matrices while eliminating the costs of manual design. AG-SpTRSV abstracts the procedures of optimizing an SpTRSV kernel as a scheme and constructs a comprehensive optimization space based on it. By defining a unified code template and preparing code variants, AG-SpTRSV enables fine-grained dynamic parallelism and adaptive code optimizations to handle various tasks. Through computation graph transformation and multi-hierarchy heuristic scheduling, AG-SpTRSV generates schemes for task partitioning and mapping, which effectively address the issues of irregular data distribution and internal data dependencies. AG-SpTRSV searches for the best scheme to optimize the target kernel for the specific matrix. A learned lightweight performance model is also introduced to reduce search costs and provide an efficient end-to-end solution. Experimental results with SuiteSparse Matrix Collection on NVIDIA Tesla A100 and RTX 3080 Ti show that AG-SpTRSV outperforms state-of-the-art implementations with geometric average speedups of 2.12x ∼ 3.99x. With the performance model enabled, AG-SpTRSV can provide an efficient end-to-end solution, with preprocessing times ranging from 3.4 to 245 times of the execution time.

Sequential node search for faster neural architecture search

Neural Architecture Search (NAS) has progressed significantly by reducing search costs since it was first introduced. The faster NAS algorithms use cell-type architectures. The cells are made up of nodes, and typically, all the node operations are searched together. This paper proposes a novel algorithm that searches the nodes of the cell type NAS independently of one another. This is done in a sequential manner, hence named SeQuential Neural Architecture Search (SQNAS). It is hypothesized that the hierarchical nature of nodes’ feature tensors is the proper guide to organize the sequence of the nodes. SQNAS reduces the size of the super graph search space at any instant of search, thus reducing the search cost considerably. The independence assumption is justified considering the competitive performance obtained. SQNAS search is also done in cycles of several restarts for better regularization. With a single NVIDIA GTX 1080Ti GPU, the search is completed in 6 h for the image classification task, obtaining a 2.55% test error on CIFAR10. For the Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) cells searched on smaller datasets, the transferability to deeper models with larger datasets is tested with ImageNet and WikiText-2. Partial channel SQNAS was also implemented, which completed the search in just 1.5 h on the CIFAR10 dataset with competitive test results. This is one of the fastest searches achieved on CIFAR10.

Neural architecture search via similarity adaptive guidance

Evolutionary neural network architecture search (ENAS) has attracted the attention of many experts due to its global optimization capabilities to automatically search for convolutional neural network architectures based on the target task. The current search space for ENAS is not to design a fully structured network, but to search for smaller cell architectures to reduce search costs. However, blind search strategies do not effectively utilize the potential experience of the population. In order to utilize the potential experience learned by the current population to guide the evolutionary search of the population, we propose a similarity guided neural network architecture search algorithm based on cell architecture, which utilizes the similarity between pairwise architectures in the population as empirical knowledge learned by the population. Our proposed algorithm provides a novel method for calculating architecture similarity, which calculates architecture similarity separately from the cell and macro-structure. Then we decouple the connections and operations in the cell and calculate connection and operation similarity separately. In addition, we propose adaptive similarity selection and binary tournament selection strategies to enhance the algorithm’s global and local search capabilities and effectively explore the search space. Finally, we design an improved single-point crossover operator to enhance the local search ability of the evolutionary operator. The experimental results show that SAGNAS is a competitive algorithm that achieves 97.44% and 81.60% in CIFAR10 and CIFAR100 with only 1.9 GPU-days spent.

The Application of Internet Technology in Enhancing the Efficiency of the Gig Economy

As our country's overall strength continues to grow, internet technology has rapidly developed, giving rise to the gig economy as a new economic form. The gig economy is characterized by its strong flexibility and intelligence, becoming the choice of many, while providing businesses with a broader range of human resource options. Traditionally, the gig economy has been limited by geographical and informational constraints, making it difficult for supply and demand to match effectively. However, with the widespread adoption of internet technology, gig workers and demanders can easily post and search for information online, achieving precise matches. This efficient circulation of information not only reduces search costs but also improves matching efficiency, making the gig economy more flexible and efficient. Additionally, internet technology enhances the efficiency of the gig economy by optimizing organizational management and processes. With tools such as project management software, automation, and intelligent processes, teams within the gig economy can better coordinate work and improve work efficiency. These technologies reduce error rates, ensure the quality and effectiveness of work, and promote the sustainable development of the gig economy.

The influence of mobile money services on customers in the Bolgatanga municipality, Ghana

Mobile telecommunications can help low-income countries improve their fixed-line and internet infrastructures. Mobile phones boost information availability, reduce search costs, facilitate supply coordination, and benefit low-income people. Mobile banking, often known as m-money, uses mobile networks to allow account balance queries and money transfers without the use of physical infrastructure. Ghanaians strongly prefer mobile money transfer services because of their low cost and accessibility across all economic strata. These simple and dependable services are suitable for clients who have little financial means, as most Ghanaians do. There has been very little empirical research into the benefits and challenges of mobile money services. Because of a paucity of data at the individual level, research into the penetration of financial services in low-income countries such as Ghana is limited. This study looks at both the advantages and disadvantages of using mobile money services. The study analyzed data from a mobile money survey using Partial Squares Structural Equation Modelling (PLS-SEM). This evidence supports the hypothesis that benefits will increase customer sentiment, while restrictions will decrease patronage. The researchers will conduct additional research into the theoretical and practical implications of mobile money systems' benefits and drawbacks for consumers. It supports the hypothesis that advantages improve consumers' perceptions, while impediments reduce patronage. Additional research will consider both the theoretical and practical implications of mobile money systems' benefits and challenges for users.

Does the proliferation of smartphones reduce consumer search costs? The case of the Korean gasoline market

AbstractThis study treats the introduction of smartphones and access to the Korean gasoline market as a natural experiment, investigating how they affect consumer search costs. We estimate consumer search costs in different regional districts in Seoul and examine how smartphone proliferation affects search costs. Our findings indicate that the widespread use of smartphones reduced consumer search costs, and this reduction was more significant in districts with higher competition across firms. We can also infer that reduced search costs narrowed the market's dispersion of prices and markups.

Navigation systems for agricultural machinery for adaptive-landscape farming

The GLONASS navigator is a versatile tool with diverse applications spanning air, sea, and land navigation. In agriculture, it facilitates adaptive-landscape farming by tailoring practices to specific soil and climate conditions, reducing search costs, and enhancing accuracy for surveyors and farmers. Beyond agriculture, GLONASS technology revolutionizes transportation by enabling shorter flight times, optimizing traffic flow, and even supporting aircraft control with autonomous landing prototypes. Its integration into various devices like watches, mobile phones, cars, and GPS-equipped dog collars further extends its utility. The widespread adoption of GLONASS in navigation and traffic management promises enhanced efficiency and safety across industries, including agriculture. As this technology evolves and becomes more accessible, its potential applications are poised to expand, offering exciting prospects for the future of adaptive-landscape farming and navigation across land, water, and air. While GLONASS is a powerful tool for navigation globally, it may face limitations in signal reception in underground areas like parking lots, caves, and chambers.

Analysis of Atom against quantum attacks

AbstractA significant amount of study is being done to review the security promises made for the various ciphers now in use as a result of the development of quantum computing technology. A general attack against symmetric key cryptography primitives that can reduce search costs to the square root is Grover's search algorithm. To implement Grover's algorithm, it is necessary that the target cipher be implemented as a quantum circuit. Despite being relatively new, this area of study has received significant attention from the research community. The authors have estimated the cost of Grover's key search attack against the stream cipher Atom, for the first time, under circuit depth restrictions defined in National Institute of Standards and Technology (NIST) PQC standardisation process. The authors implement the quantum circuit of Atom in QISKIT, (open‐source software development kit for working with quantum computers running on IBM Quantum Experience). The results are also compared with other existing literature on LFSR‐based stream ciphers, such as Grain‐v1, Grain‐128‐AEAD, and Lizard. The authors also find that, to the best of their knowledge, in the existing literature on estimating the cost of Grover's attack on symmetric ciphers, Atom is the only 128‐bit key cipher that meets the threshold of ≈2170 set by NIST for quantum security of 128‐bit key ciphers. The authors also analyse the security of Atom against quantum TMDTO attacks.

Trademarks in the Age of Automated Commerce: Consumer Choice and Autonomy

In the age of automated commerce, powered by artificial intelligence and machine learning (AI/ML) tools, the role of trademarks in shopping may be diminishing. For instance, the Alexa voice assistant will announce only a couple of purchase options under specific trademarks, concealing the plenitude of other products on its interface. Another example is that Amazon will pursue the shipping-then-shopping scenario, with trademarks being perceived by consumers only upon delivery. Moreover, it has been predicted that, in some cases, the new AI/ML tools will search for products irrespective of trademarks and will do so faster than any human being. Under those circumstances, consumer search costs will be lower, not because of trademarks but because of the new shopping architecture based on those tools. While lowering consumer search costs has traditionally been the role of trademarks, the availability of other tools for the same purpose may be a positive development. However, another trend is that the new AI/ML tools are constantly taking part in consumer decision-making, possibly reducing consumers’ freedom of choice and personal autonomy. In attempting to tackle this issue, it is worth approaching trademarks from the perspective of choice and autonomy. Seen from this angle, it may be possible to learn specific lessons from European Union (EU) trademark law that will assist in reshaping the automated – and autonomous – shopping architecture for the benefit of consumer well-being. In this exercise, close attention will be paid to the case-law of the Court of Justice of the European Union, with its discourse on product alternatives, while due regard will be shown to the new legislation on the platform economy in the EU. In the end, this paper will demonstrate that trademarks are valuable not only because they help reduce search costs but also because they promote alternatives and, thus, improve consumers’ general autonomy.

Do Lower Search Costs Benefit Intermediaries?

Intermediaries—such as Amazon, Alibaba Group, and Walmart—play critical roles in digital markets: This paper proposes a model for analyzing how search cost affects their market outcomes. Distinguished from existing search literature, we introduce network effects to the market by allowing buyers to have positive preferences for product variety. In addition, we relax the exogenous demand and supply assumptions in the conventional search literature to study how search affects participation on both sides of the market. Through the model, we focus on studying and comparing two important types of intermediary: the two-sided platform; and the one-sided retailer. We find that reducing search costs does not necessarily increase the profit of both intermediary types. Furthermore, its influences are highly varied—depending on both the type of intermediary and its buyers’ preferences for variety.

The Voice of Commerce: How Smart Speakers Reshape Digital Content Consumption and Preferences

The present study investigates the effects of smart speaker usage on consumers’ digital content search, purchase, and consumption behaviors. Using a unique panel data set comprising information on household patterns of digital content (e.g., video on demand [VOD]) transactions and consumption and smart speaker usage, we found that the adoption of smart speakers is positively associated with the increased purchase of digital content but negatively related to the average rate of content completion. More specifically, we found that VOD content-related expenditures increased by 21.5% following smart speaker adoption but the average consumption of VOD content purchased decreased by 3.0%. We also examined millions of data points on TV remote-control use and conducted a survey via MTurk to support the validity of the findings. Smart speaker usage can reduce search costs, which subsequently increases search incidence and conversion rates, behavioral changes that can lead to a rise in purchases. We further show that the use of smart speakers for purposes other than information seeking is positively associated with purchases. We develop insights on how to elicit economic value from voice recognition technologies and provide implications for the design and implementation of effective voice commerce strategies.

DE-DARTS: Neural architecture search with dynamic exploration

Neural architecture search (NAS) methods automatically find optimal neural networks without human assistance. Numerous algorithms for NAS have been studied to find architectures with gradient-based search. Differentiable architecture search (DARTS), one of the key papers of gradient-based search, dramatically reduced search cost, and showed outstanding performance through continuous relaxation and meta-learning based approximation. However, one of the issues with DARTS is that the gradient-based search process is biased due to the nested bi-level optimization structure, and the greedy behavior of the gradient descent. As a result, there is a problem that search spaces are limited to a limited set of architectures. To overcome the bias of the gradient-based search in network architecture search (NAS), we used a dynamic search method. This technique allows gradient-based search to have exploration. In this paper, we present a novel approach, namely, Dynamic-Exploration DARTS (DE-DARTS). For effective exploration, we use dynamic attention networks (DANs) in DE-DARTS, which change model architectures based on input data. As our DANs are activated early in the search, more diverse architectures are considered, depending on input data at the beginning of search. Our algorithm is evaluated in multiple image classification datasets including CIFAR-10, and ImageNet, and shows improved performance.

Learning by searching: Spatial mismatches and imperfect information in Southern labor markets

Youth unemployment remains extremely high throughout the developing world, at times coexisting with unmet demand for labour and high job turnover. We examine one possible explanation for this: spatial mismatches between jobs and job-seekers combined with high search costs can lead young job-seekers to have overly optimistic beliefs about their employment prospects. As a result, job-seekers under-search but also hold out for better jobs. Through a field experiment we find that reducing search costs through transport subsidies leads job-seekers to search more intensively and to adjust their beliefs in line with their search experience. When jobs fail to materialize immediately, job-seekers who believed that dropping CVs at prospective employers in the city centre was an effective search strategy become more impatient, they lower their reservation wage and they settle for low-paying jobs closer to home. This does not increase their likelihood of being employed, since nearby jobs are also scarce. These findings underscore both the importance and the complexity of the interaction between search costs and beliefs, and how they can lead to spatial and occupational mistargeting in the job search.

A Recommendation Mechanism for a Non-Fungible Token Trading Platform

Technology has evolved at an increasing pace over the decades, blockchain is used in many fields, and Non-Fungible Token (NFT) is one of its applications. With NFT, all digital assets can be freely traded because NFT establishes a unique token for each asset. From the transaction statistics, it can be found that the transaction volume in the past six months has increased explosively, and the projects uploaded to the NFT trading platform have also increased; but it also means customers need to spend more time finding what they want. Active recommendation systems can be used to solve this problem to reduce search cost. This study proposes a recommendation system for NFT tokens, compares the similarity between items, and analyzes buyers' collection data to prove that the recommendation theory proposed in this study is valid.

Strategic referrals and on-the-job search equilibrium

Referrals are prevalent in the U.S. labor market. To understand their aggregate effects, this paper studies an equilibrium model of on-the-job search and job referrals. In the model, referrals are modeled as a strategic interaction between a referrer and a firm. The equilibrium model shows that referrals benefit job searchers whose outside option is above a threshold. I support this prediction by showing that the referral wage premium exists only for employed job searchers. Quantitatively, referrals contribute to the total output by 3.93% through transmitting information and reducing search costs. The information transmission explains about 28% of the effects.

Death bells? Branding in the age of smartphones and AI

Purpose Branding has arguably been the most crucial marketing pillar in the twentieth century. It was effective because of existing consumer behavior, which was constrained by the availability of information and customers’ ability to process it, resulting in a reliance on brands. This paper aims to examine the role of branding (and brand loyalty) in the past and how it has been disrupted recently, and makes recommendations for practicing managers to modify how they manage brands proactively. Design/methodology/approach This work is based on a review of the latest developments in the theory and practice of branding. Findings Today, ready access to smartphones ensures the availability of information tailored to customer needs, directing them in making choices. Improvements in the quality of artificial intelligence (AI)-driven algorithms further simplify customer purchase decisions and reduce search costs – a blow to branding. The data, which forms the foundation for information sought by customers and AI algorithms, continues to increase as more buyers leave their digital footprints, resulting in a virtuous self-precipitating cycle of better decision-making for customers and compromising the influence of brands. Originality/value To the best of authors’ knowledge, this is the first paper to focus on the implications of changes in the marketplace driven by smartphones and AI on the future of branding.