ID Space Research Articles

MMVFL: A Simple Vertical Federated Learning Framework for Multi-Class Multi-Participant Scenarios.

Federated learning (FL) is a privacy-preserving collective machine learning paradigm. Vertical federated learning (VFL) deals with the case where participants share the same sample ID space but have different feature spaces, while label information is owned by one participant. Early studies of VFL supported two participants and focused on binary-class logistic regression problems, while recent studies have put more attention on specific aspects such as communication efficiency and data security. In this paper, we propose the multi-participant multi-class vertical federated learning (MMVFL) framework for multi-class VFL problems involving multiple parties. By extending the idea of multi-view learning (MVL), MMVFL enables label sharing from its owner to other VFL participants in a privacy-preserving manner. To demonstrate the effectiveness of MMVFL, a feature selection scheme is incorporated into MMVFL to compare its performance against supervised feature selection and MVL-based approaches. The proposed framework is capable of quantifying feature importance and measuring participant contributions. It is also simple and easy to combine with other communication and security techniques. The experiment results on feature selection for classification tasks on real-world datasets show that MMVFL can effectively share label information among multiple VFL participants and match the multi-class classification performance of existing approaches.

Flexible Vertical Federated Learning With Heterogeneous Parties.

We propose flexible vertical federated learning (Flex-VFL), a distributed machine algorithm that trains a smooth, nonconvex function in a distributed system with vertically partitioned data. We consider a system with several parties that wish to collaboratively learn a global function. Each party holds a local dataset; the datasets have different features but share the same sample ID space. The parties are heterogeneous in nature: the parties' operating speeds, local model architectures, and optimizers may be different from one another and, further, they may change over time. To train a global model in such a system, Flex-VFL utilizes a form of parallel block coordinate descent (P-BCD), where parties train a partition of the global model via stochastic coordinate descent. We provide theoretical convergence analysis for Flex-VFL and show that the convergence rate is constrained by the party speeds and local optimizer parameters. We apply this analysis and extend our algorithm to adapt party learning rates in response to changing speeds and local optimizer parameters. Finally, we compare the convergence time of Flex-VFL against synchronous and asynchronous VFL algorithms, as well as illustrate the effectiveness of our adaptive extension.

EFMVFL: An Efficient and Flexible Multi-party Vertical Federated Learning without a Third Party

Federated learning (FL) is a machine learning setting which allows multiple participants collaboratively to train a model under the orchestration of a server without disclosing their local data. Vertical federated learning (VFL) is a special structure in FL. It handles the situation where participants have the same ID space but different feature spaces. In order to guarantee the security and privacy of the local data of each participant, homomorphic encryption (HE) is often used to transmit intermediate parameters or data during the training process. In most VFL frameworks, a trusted third-party server is necessary because the plaintexts of the parameters need to be revealed for the computation. However, it is hard to find such a credible entity in the real world. Existing methods for solving this problem are either communication-intensive or unsuitable for multi-party scenarios. By combining secret sharing (SS) and HE, we propose a novel VFL framework without any trusted third parties called EFMVFL. It allows intermediate parameters to be transmitted among multiple parties without revealing the plaintexts. EFMVFL is applicable to generalized linear models (GLMs) and supports flexible expansion to multiple participants. Extensive experiments under Logistic Regression and Poisson Regression show that our framework is outstanding in communication (reduced by 3.2×– 6.8×) and efficiency (accelerated by 1.6×– 3.1×).

RFID-Based IoT Scanning System for Tracking Parking Slot Application

The increasing demand for parking spaces in various facilities, such as shopping centers, ports, and airports, poses a significant challenge for parking managers. Drivers waste time and fuel searching for available parking slots, resulting in increased air pollution and negative environmental and health impacts. Therefore, a new solution is needed to address these issues and improve parking management efficiency. A parking management system that utilizes RFID technology and allows for the RFID Tag on each car to display user data, user ID, and parking space reservations is needed to improve the efficiency and convenience of parking at various facilities. The proposed system design incorporates various components, including a Camera Sensor, Raspberry PI, ESP8266, UHF RFID reader, and RFID Tag. The RFID technology captures radio waves emitted by the RFID Tag, enabling individual identification through unique codes. Testing revealed high accuracy at close range but decreased effectiveness at longer distances. Internet speed and network latency affected system performance, with faster speeds enabling faster data transmission. The system, developed using Thunkable and Firebase, provides real-time parking space availability information and streamlines vehicle entry and exit processes through RFID technology. This system has the potential to greatly improve the parking experience at busy public facilities by facilitating efficient parking spot finding and enhancing overall convenience for drivers.

Mining Distributed Data using Vertical Federated Learning Review


 
 
 
 Federated Learning was designed to allow collaborative learning without revealing raw data as worries about machine learning privacy grew. Vertical Federated Learning (VFL) may be utilized for a distributed dataset with the same sample ID space but differs in feature space. And may be used in a wide variety of real-world contexts when parties have the same set of samples but only have partial attributes. Achieving privacy will be a result of this technique's capacity.
 Federated Learning enables different repositories of data to learn a shared model collaboratively and at the same time keep the privacy of each one because of the increasing awareness of large firms compromising on data security and user privacy. To accomplish federated learning, three learning ways were suggested; horizontal federated learning, vertical federated learning, and transfer federated learning.
 Vertical federated learning was adopted when data were spread among different parties. However, each one has different features from the others for identical objects. This paper is related to this type of federated learning.
 
 
 
 
 To maximize model performance while maintaining the privacy of dispersed data, we'll create a framework based on vertical federated learning and suitable techniques.

Vertically Federated Learning with Correlated Differential Privacy

Federated learning (FL) aims to address the challenges of data silos and privacy protection in artificial intelligence. Vertically federated learning (VFL) with independent feature spaces and overlapping ID spaces can capture more knowledge and facilitate model learning. However, VFL has both privacy and utility problems in framework construction. On the one hand, sharing gradients may cause privacy leakage. On the other hand, the increase in participants brings a surge in the feature dimension of the global model, which results in higher computation costs and lower model accuracy. To address these issues, we propose a vertically federated learning algorithm with correlated differential privacy (CRDP-FL) to meet FL systems’ privacy and utility requirements. A privacy-preserved VFL framework is designed based on differential privacy (DP) between organizations with many network edge devices. Meanwhile, feature selection is performed to improve the algorithm’s efficiency and model performance to solve the problem of dimensionality explosion. We also propose a quantitative correlation analysis technique for VFL to reduce the correlated sensitivity and noise injection, balancing the utility decline due to DP protection. We theoretically analyze the privacy level and utility of CRDP-FL. A real vertically federated learning scenario is simulated with personalized settings based on the ISOLET and Breast Cancer datasets to verify the method’s effectiveness in model accuracy, privacy budget, and data correlation.

Semi-Supervised Federated Heterogeneous Transfer Learning

Federated learning (FL) is a privacy-preserving paradigm that collaboratively train machine learning models with distributed data stored in different silos without exposing sensitive information. Different from most existing FL approaches requiring data from different parties share either the same feature space or sample ID space, federated transfer learning (FTL), which is a recently proposed FL concept, is designed for situations where data from different parties differ not only in samples but also in feature space. However, like most traditional FL approaches, FTL methods also suffer from issues caused by insufficiency of overlapping data. In this paper, we propose a novel FTL framework referred to as Semi-Supervised Federated Heterogeneous Transfer Learning (SFHTL) to leverage on the unlabeled non-overlapping samples to reduce model overfitting as a result of insufficient overlapping training samples in FL scenarios. Unlike existing FTL approaches, SFHTL makes use of non-overlapping samples from all parties to expand the training set for each party to improve local model performance. Through extensive experimental evaluation based on real-world datasets, we demonstrate significant advantages of SFHTL over state-of-the-art approaches.

CANDIDATE: A tool for generating anonymous participant-linking IDs in multi-session studies.

To ensure the privacy of participants is an ethical and legal obligation for researchers. Yet, achieving anonymity can be technically difficult. When observing participants over time one needs mechanisms to link the data from the different sessions. Also, it is often necessary to expand the sample of participants during a project. To help researchers simplify the administration of such studies the CANDIDATE tool is proposed. This tool allows simple, unique, and anonymous participant IDs to be generated on the fly. Simulations were used to validate the uniqueness of the IDs as well as their anonymity. The tool can successfully generate IDs with a low collision rate while maintaining high anonymity. A practical compromise between integrity and anonymity was achieved when the ID space is about ten times the number of participants. The tool holds potential for making it easier to collect more comprehensive empirical evidence over time that in turn will provide a more solid basis for drawing reliable conclusions based on research data. An open-source implementation of the tool that runs locally in a web-browser is made available.

The CSI Framework for Compiler-Inserted Program Instrumentation

The CSI framework provides comprehensive static instrumentation that a compiler can insert into a program-under-test so that dynamic-analysis tools - memory checkers, race detectors, cache simulators, performance profilers, code-coverage analyzers, etc. - can observe and investigate runtime behavior. Heretofore, tools based on compiler instrumentation would each separately modify the compiler to insert their own instrumentation. In contrast, CSI inserts a standard collection of instrumentation hooks into the program-under-test. Each CSI-tool is implemented as a library that defines relevant hooks, and the remaining hooks are "nulled" out and elided during either compile-time or link-time optimization, resulting in instrumented runtimes on par with custom instrumentation. CSI allows many compiler-based tools to be written as simple libraries without modifying the compiler, lowering the bar for the development of dynamic-analysis tools. We have defined a standard API for CSI and modified LLVM to insert CSI hooks into the compiler's internal representation (IR) of the program. The API organizes IR objects - such as functions, basic blocks, and memory accesses - into flat and compact ID spaces, which not only simplifies the building of tools, but surprisingly enables faster maintenance of IR-object data than do traditional hash tables. CSI hooks contain a "property" parameter that allows tools to customize behavior based on static information without introducing overhead. CSI provides "forensic" tables that tools can use to associate IR objects with source-code locations and to relate IR objects to each other. To evaluate the efficacy of CSI, we implemented six demonstration CSI-tools. One of our studies shows that compiling with CSI and linking with the "null" CSI-tool produces a tool-instrumented executable that is as fast as the original uninstrumented code. Another study, using a CSI port of Google's ThreadSanitizer, shows that the CSI-tool rivals the performance of Google's custom compiler-based implementation. All other demonstration CSI tools slow down the execution of the program-under-test by less than 70%.

The CSI Framework for Compiler-Inserted Program Instrumentation

The CSI framework provides comprehensive static instrumentation that a compiler can insert into a program-under-test so that dynamic-analysis tools - memory checkers, race detectors, cache simulators, performance profilers, code-coverage analyzers, etc. - can observe and investigate runtime behavior. Heretofore, tools based on compiler instrumentation would each separately modify the compiler to insert their own instrumentation. In contrast, CSI inserts a standard collection of instrumentation hooks into the program-under-test. Each CSI-tool is implemented as a library that defines relevant hooks, and the remaining hooks are "nulled" out and elided during either compile-time or link-time optimization, resulting in instrumented runtimes on par with custom instrumentation. CSI allows many compiler-based tools to be written as simple libraries without modifying the compiler, lowering the bar for the development of dynamic-analysis tools. We have defined a standard API for CSI and modified LLVM to insert CSI hooks into the compiler's internal representation (IR) of the program. The API organizes IR objects - such as functions, basic blocks, and memory accesses - into flat and compact ID spaces, which not only simplifies the building of tools, but surprisingly enables faster maintenance of IR-object data than do traditional hash tables. CSI hooks contain a "property" parameter that allows tools to customize behavior based on static information without introducing overhead. CSI provides "forensic" tables that tools can use to associate IR objects with source-code locations and to relate IR objects to each other. To evaluate the efficacy of CSI, we implemented six demonstration CSI-tools. One of our studies shows that compiling with CSI and linking with the "null" CSI-tool produces a tool-instrumented executable that is as fast as the original uninstrumented code. Another study, using a CSI port of Google's ThreadSanitizer, shows that the CSI-tool rivals the performance of Google's custom compiler-based implementation. All other demonstration CSI tools slow down the execution of the program-under-test by less than 70%.

IPersea: Towards improving the Sybil-resilience of social DHT

P2P systems are highly susceptible to Sybil attacks, in which an attacker creates a large number of identities and uses them to control a substantial fraction of the system. Persea is the most recent approach towards designing a social network based Sybil-resistant DHT. Unlike prior Sybil-resistant P2P systems based on social networks, Persea does not rely on two key assumptions: (i) that the social network is fast mixing, and (ii) that there is a small ratio of attack edges to honest peers. Both assumptions have been shown to be unreliable in real social networks. The hierarchical distribution of node IDs in Persea confines a large attacker botnet to a considerably smaller region of the ID space than in a normal P2P system and its replication mechanism lets a peer to retrieve the desired results even if a given region is occupied by attackers. However, Persea system suffers from certain limitations, since it cannot handle the scenario, where the malicious target returns an incorrect result instead of just ignoring the lookup request. In this paper, we address this major limitation of Persea through a Sybil detection mechanism built on top of Persea system, which accommodates inspection lookup, a specially designed lookup scheme to detect the Sybil nodes based on their responses to the lookup query. We design a scheme to filter those detected Sybils to ensure the participation of honest nodes on the lookup path during regular DHT lookup. Since the malicious nodes are opt-out from the lookup path in our system, they cannot return any incorrect result during regular lookup. We evaluate our system in simulations with social network datasets and the results show that catster, the largest network in our simulation with 149,700 nodes and 5,449,275 edges, gains 100% lookup success rate, even when the number of attack edges is equal to the number of benign peers in the network.

Probabilistic Optimal Tree Hopping for RFID Identification

Radio frequency identification (RFID) systems are widely used in various applications such as supply chain management, inventory control, and object tracking. Identifying RFID tags in a given tag population is the most fundamental operation in RFID systems. While the Tree Walking (TW) protocol has become the industrial standard for identifying RFID tags, little is known about the mathematical nature of this protocol, and only some ad hoc heuristics exist for optimizing it. In this paper, first we analytically model the TW protocol, and then using that model, propose the Tree Hopping (TH) protocol that optimizes TW both theoretically and practically. The key novelty of TH is to formulate tag identification as an optimization problem and find the optimal solution that ensures the minimal average number of queries or identification time as per the requirement. With this solid theoretical underpinning, for different tag population sizes ranging from 100 to 100 K tags, TH significantly outperforms the best prior tag identification protocols on the metrics of the total number of queries per tag, the total identification time per tag, and the average number of responses per tag by an average of 40%, 59%, and 67%, respectively, when tag IDs are nonuniformly distributed in the ID space, and of 50%, 10%, and 30%, respectively, when tag IDs are uniformly distributed.

A SCALABLE SECURITY SERVICE FOR GEOGRAPHIC AD-HOC ROUTING

This paper describes a scalable security service for geographic ad-hoc routing. The routing protocol, REWARD, detects black hole attacks and organizes a distributed data base for suspicious nodes and areas. The algorithm utilizes two types of broadcast messages, MISS and SAMBA, to recruit nodes to act as security servers. Security servers keep records for detected black hole attacks and provide security services when forward packets. MISS-recruited security servers keep records for suspicious nodes and protect the network in the ID space. SAMBArecruited security servers keep records for suspicious areas and decline the network vulnerability in the physical space. REWARD has different levels of security which can be set according to the local conditions. In order to determine the effectiveness of REWARD we used ANTS, a simulation environment which models the traffic of wireless sensor networks.

An Efficient and Scalable Routing for MANETs

Recently, Distributed-Hash-Table (DHT)-based routing protocols have been proposed for large scale mobile ad hoc networks (MANETs). The motive behind using DHT is to avoid/reduce network wide flooding in MANETs in route discovery phase leading to achieve more scalable network. The challenge of deploying DHT at network layer for routing purpose in MANETs is to achieve more matching between logical and physical networks. The paper describes that existing DHT-based routing protocols for MANETs lack this feature. The paper presents a new DHT based scheme [called an efficient and scalable routing for MANETs (ESR)] which reduces the ill-matching between logical and physical networks. This is achieved be distributing the logical identifier (LID) space of DHT structure among the nodes such that the physical neighbors have consecutive LID space portions, i.e. all physical neighbors of a node are also the logical neighbors of the node. Therefore the logical ID space portion of a node may be non-contiguous. Based on this logical structure, a node builds up binary-search-tree (BST) using both logical ID space portion of itself and its neighbors. This BST at the node is traversed to find the next hop for a query/message. Through simulation, the paper shows the proposed approach (ESR) performs better than the existing one in term of routing overhead, average end-to-end delay, path-stretch values and false-negative ratio.

Reliability Analysis of Structured P2P System

Reliability is most important for structured p2p systems and how to exactly evaluate the metric-loss rate which is tightly related to reliability is a hot topic and challenge. Many research works of the loss rate only consider the situation that the loss rate is caused by next hop node failure (NF). Indeed, the lost of queuing message (QF) on the failure node also contribute to the loss rate. This paper presents an analytical model of loss rate caused by NF and NQ. In order to achieve the optimal performance of system through minimizing the NQ, we propose two methods: process power enhancing (PPE) and Space Partition (SP). In the heterogeneous systems, we can obtain the minimal loss rate via improving each nodes process power; in the homogeneous structured p2p systems, the partition space method can evenly partition the ID space so as to ensure the traffic load is uniformly distributed over all nodes.

Probabilistic optimal tree hopping for RFID identification

Radio Frequency Identification (RFID) systems are widely used in various applications such as supply chain management, inventory control, and object tracking. Identifying RFID tags in a given tag population is the most fundamental operation in RFID systems. While the Tree Walking (TW) protocol has become the industrial standard for identifying RFID tags, little is known about the mathematical nature of this protocol and only some ad-hoc heuristics exist for optimizing it. In this paper, first, we analytically model the TW protocol, and then using that model, propose the Tree Hopping (TH) protocol that optimizes TW both theoretically and practically. The key novelty of TH is to formulate tag identification as an optimization problem and find the optimal solution that ensures the minimal average number of queries. With this solid theoretical underpinning, for different tag population sizes ranging from 100 to 100K tags, TH significantly outperforms the best prior tag identification protocols on the metrics of the total number of queries per tag, the total identification time per tag, and the average number of responses per tag by an average of 50%, 10%, and 30%, respectively, when tag IDs are uniformly distributed in the ID space, and of 26%, 37%, and 26%, respectively, when tag IDs are non-uniformly distributed.

Leader election using loneliness detection

We consider the problem of leader election (LE) in single-hop radio networks with synchronized time slots for transmitting and receiving messages. We assume that the actual number n of processes is unknown, while the size u of the ID space is known, but is possibly much larger. We consider two types of collision detection: strong (SCD), whereby all processes detect collisions, and weak (WCD), whereby only non-transmitting processes detect collisions. We introduce loneliness detection (LD) as a key subproblem for solving LE in WCD systems. LD informs all processes whether the system contains exactly one process or more than one. We show that LD captures the difference in power between SCD and WCD, by providing an implementation of SCD over WCD and LD. We present two algorithms that solve deterministic and probabilistic LD in WCD systems with time costs of $${\mathcal{O}(\log \frac{u}{n})}$$ and $${\mathcal{O}(\min( \log \frac{u}{n}, \frac{\log (1/\epsilon)}{n}))}$$ , respectively, where $${\epsilon}$$ is the error probability. We also provide matching lower bounds. Assuming LD is solved, we show that SCD systems can be emulated in WCD systems with factor-2 overhead in time. We present two algorithms that solve deterministic and probabilistic LE in SCD systems with time costs of $${\mathcal{O}(\log u)}$$ and $${\mathcal{O}(\min ( \log u, \log \log n + \log (\frac{1}{\epsilon})))}$$ , respectively, where $${\epsilon}$$ is the error probability. We provide matching lower bounds.

Heuristic Query Tree Protocol: Use of Known Tags for RFID Tag Anti-Collision

†SUMMARY Existing query tree protocols deal with RFID tags in a blind manner. They query tags in a fixed bit order based on the assumption that the tag ID numbers are uniformly distributed throughout the range of the entire ID space because readers have no prior knowledge of the tags. This paper attempts to distinguish RFID applications where readers are already aware of all tags used by the application. We propose a heuristic query tree (H-QT) protocol that uses heuristic to select effective bits from known tags for the best queries in a divide and conquer approach. The performance evaluation shows that the proposed protocol is superior to original query tree protocols because it significantly reduces the number of tag collisions and no tag response.

Indexing Associated Knowledge Flow on the Web

The Associated Knowledge Flow (AKF) on the Web is an ordered sequence of Web pages that have associated relation. The associated relation from page A to page B indicates that users who have browsed page A is likely to also browse page B. The motivation of this paper is to index the AKFs on the Web and provide users AKFs instead of discrete resources. We build a scalable P2P-based Web resource-sharing system and design two kinds of ID spaces (hash ID space and semantic ID space) on it to index resources and facilitate AKF discovery. Theoretical analysis and simulations show that such a system can achieve logarithmic performance and cost.

How do you 'Tube'

In this paper we "reverse-engineer" the YouTube video delivery cloud by building a distributed measurement infrastructure. Through extensive data collection and analysis, we deduce the key design features underlying the YouTube video delivery cloud. The design of the YouTube video delivery cloud consists of three major components: a "flat" video id space, multiple DNS namespaces reflecting a multi-layered logical organization of video servers, and a 3-tier physical cache hierarchy. By mapping the video id space to the logical servers via consistent hashing and cleverly leveraging DNS and HTTP re-direction mechanisms, such a design leads to a scalable, robust and flexible content distribution system.