303 research outputs found
Money Talks? An Experimental Study of Rebate in Reputation System Design
No full textReputation systems that rely on feedback from traders are important institutions for helping sustain trust in markets, while feedback information is usually considered a public good. We apply both theoretical models and experiments to study how raters' feedback behavior responds to different reporting costs and how to improve market efficiency by introducing a pre-commitment device for sellers in reputation systems. In particular, the pre-commitment device we study here allows sellers to provide rebates to cover buyers' reporting costs before buyers make purchasing decisions. Using a buyer-seller trust game with a unilateral feedback scheme, we find that a buyer’s propensity to leave feedback is more sensitive to reporting costs when the seller cooperates than when the seller defects. The seller’s decision on whether to provide a rebate significantly affects the buyer’s decision to leave feedback by compensating for the feedback costs. More importantly, the rebate decision has a significant impact on the buyer's purchasing decision via signaling the seller's cooperative type. The experimental results show that the rebate mechanism improves the market efficiency.reputation, trust, feedback mechanism, asymmetric information, public goods, experimental economics
Topological edge states in one-dimensional non-Hermitian Su-Schrieffer-Heeger systems of finite lattice size: Analytical solutions and exceptional points
No full textAutomatic video genre categorization and event detection techniques on large-scale sports data
No full textThis paper presents an efficient and robust automatic process for large-scale sports video analysis. The proposed system firstly identifies the genre of the query video, and then accomplishes the interesting event detection task. The significance of this framework is its automatic characteristic in testing with minimum human involvement in training, as well as the scalability and expansibility in dealing with a large-scale dataset. Domain-knowledge independent local features are extracted from an input video sequence and a histogram based distribution representation is created using the bag-of-visual-words (BoW) model. In genre categorization, k-nearest neighbor (k-NN) classifiers with various dissimilarity measures are assessed and evaluated analytically. For the event detection, a hidden conditional random field (HCRF) structured prediction model is utilized. Overall, this framework demonstrates the efficiency and accuracy in processing voluminous data from sports collection and achieves various tasks in video analysis. It also demonstrates a potential technology transformation from the 'laboratory bench' to commercial applications. ? 2010 Ning Zhang, Ling-Yu Duan, Qing-ming Huang, Lingfang Li, Wen Gao and Ling Guan.EI
A Novel Composite Li3V2(PO4)3∥Li2NaV2(PO4)3/C as Cathode Material for Li-Ion Batteries
No full text
A novel composite cathode for lithium ion batteries, Li3V2(PO4)3‖Li2NaV2(PO4)3/C, was synthesized by a sol-gel method. Cetyltrimethylammonium bromide (CTAB) was used as a surfactant while polyvinylidene difluoride (PVDF) was the carbon source. X-ray diffraction (XRD) and Raman results showed that the components of this composite are monoclinic Li3V2(PO4)3, rhombohedral Li2NaV2(PO4)3 and an amorphous carbon-coating. Four potential plateaus occur at the charge/discharge curves and the longest plateau is observed at a potential of 3.8/3.7 V. Therefore, the alkali metal ion intercalation and deintercalation mostly occur at this potential, which is different to that observed for Li3V2(PO4)3. In addition to the stable working potential, this composite also possesses an outstanding electrochemical performance. The sample containing 8.32 % carbon content delivers a capacity of 119 mAh g−1 at 0.2 C rate and 87 mAh g−1 at 12 C. After 50 charge/discharge cycles at 1 C, a coulombic efficiency of 98.4 % is maintained. This enhancement of the electrochemical performance could be attributed to the synergistic effect between monoclinic Li3V2(PO4)3 and rhombohedral Li2NaV2(PO4)3.
</jats:p
Bonding condition monitoring and evaluation of CFRP strengthened RC structures with piezoelectric sensors
No full textDecision Making Using Rating Systems: When Scale Meets Binary
No full textRating systems measuring quality of products and services (i.e., the state of the world) are
widely used to solve the asymmetric information problem in markets. Decision makers typically
make binary decisions such as buy/hold/sell based on aggregated individuals' opinions presented
in the form of ratings. Problems arise, however, when different rating metrics and aggregation
procedures translate the same underlying popular opinion to different conclusions about the
true state of the world. This paper investigates the inconsistency problem by examining the
mathematical structure of the metrics and their relationship to the aggregation rules. It is
shown that at the individual level, the only scale metric (1,. . . ,N) that reports people's opinion
equivalently in the a binary metric (-1, 0, 1) is one where N is odd and N-1 is not divisible by
4. At aggregation level, however, the inconsistencies persist regardless of which scale metric is
used. In addition, this paper provides simple tools to determine whether the binary and scale
rating systems report the same information at individual level, as well as when the systems
di®er at the aggregation level
Dc Line-Interactive Uninterruptible Power Supply (UPS) with Load Leveling for Constant Power and Pulse Loads
No full textUninterruptable Power Supply (UPS) systems are usually considered as a backup power for electrical systems, providing emergency power when the main power source fails. UPS systems ensure an uninterruptible, reliable and high quality electrical power for systems with critical loads in which a continuous and reliable power supply is a vital requirement. A novel UPS system topology, DC line-interactive UPS, has been introduced. The new proposed UPS system is based on the DC concept where the power flow in the system has DC characteristic. The new DC UPS system has several advantageous with respect to the on-line 3-phase UPS which is extensively used in industry, such as lower size, cost and weight due to replacing the three-phase dual converter in the on-line UPS system with a single stage single phase DC/DC converter and thus higher efficiency is expected. The proposed system will also provide load leveling feature for the main AC/DC rectifier which has not been offered by conventional AC UPS systems. It applies load power smoothing to reduce the rating of the incoming AC line and consequently reduce the installation cost and time. Moreover, the new UPS technology improves the medical imaging system up-time, reliability, efficiency, and cost, and is applicable to several imaging modalities such as CT, MR and X-ray as well. A comprehensive investigation on different energy storage systems was conducted and couple of most promising Li-ion cell chemistries, LFP and NCA types, were chosen for further aggressive tests. A battery pack based on the LFP cells with monitoring system was developed to be used with the DC UPS testbed. The performance of the DC UPS has also been investigated. The mathematical models of the system are extracted while loaded with constant power load (CPL) and constant voltage load (CVL) during all four modes of operation. Transfer functions of required outputs versus inputs were extracted and their related stability region based on the Routh-Hurwitz stability criteria were found. The AC/DC rectifier was controlled independently due to the system configuration. Two different control techniques were proposed to control the DC/DC converter. A linear dual-loop control (DLC) scheme and a nonlinear robust control, a constant frequency sliding mode control (CFSMC) were investigated. The DLC performance was convincing, however the controller has a limited stability region due to the linearization process and negative incremental impedance characteristics of the CPL which challenges the stability of the system. A constant switching frequency SMC was also developed based on the DC UPS system and the performance of the system were presented during different operational modes. Transients during mode transfers were simulated and results were depicted. The controller performances met the control goals of the system. The voltage drop during mode transitions, was less than 2% of the rated output voltage. Finally, the experimental results were presented. The high current discharge tests on each selected Li-ion cell were performed and results presented. A testbed was developed to verify the DC UPS system concept. The test results were presented and verified the proposed concept
Unraveling the Influence of Training Data and Internal Structures in Large Language Models for Enhanced Explainability (Student Abstract)
No full textRecent advances in deep learning have expanded the application of large language models (LLMs) across fields such as medicine, finance, and education. Understanding the mechanisms underlying these models is essential to mitigate issues like hallucinations and bias. This study provides deep learning practitioners with insights into how specific training data points and internal structures influence model behaviour. Using influence functions and mechanistic interpretability, we will analyze the impact of data on model predictions across various tasks. Preliminary findings indicate that semantic search techniques, such as FAISS, enable efficient identification of influential training points in GPT-2 small. Future work will extend these methods to additional tasks and more complex models, with a focus on further elucidating LLM structures to improve interpretability
- …
