Sinem Sav
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sinem Sav has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Abdullah Ercument Cicek
Department of Computer Engineering, Bilkent University, Ankara, Türkiye
4 shared publications - Jean-Pierre Hubaux
EPFL, Lausanne, Switzerland
4 shared publications - Alptekin Küpçü
Koc University
3 shared publications - Dimitrios Bounias
German Cancer Research Center (DKFZ) Heidelberg, Division of Medical Image Computing, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany; Medical Faculty Heidelberg, Heidelberg University, Im Neuenheimer Feld 672, 69120 Heidelberg, Germany
3 shared publications - Lena Maier-Hein
German Cancer Research Center
3 shared publications - Ralf Floca
Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany
3 shared publications
Latest Publications
- Open Access
- 11 September 2026
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 14, e88390
Privacy Leakage in Federated Learning in Radiology Reports: Comparative Evaluation of Tokenizer and Batch-Size Privacy Risks
Background: Federated learning (FL) enables multi-institutional model training on clinical text without sharing raw data; however, gradient inversion methods can reconstruct sensitive information from shared model updates. The extent of such privacy leakage in FL applied to radiology reports, and the role of tokenizer design, remains unclear. Objective: This study aimed to quantify gradient-based reconstruction of radiology report text in an FL setting and to compare privacy risk across 3 transformer tokenization strategies in a controlled, tokenizer-aware evaluation. Methods: Six FL clients trained a GPT-2–style transformer (sequence length 32) on 2 public clinical-text corpora comprising 368,751 diagnostic reports, 98,206 discharge summaries, and 1500 MIMIC-CXR (Medical Information Mart for Intensive Care Chest X-Ray) radiology reports. Models were trained using 3 tokenizers (GPT-2, RadBERT, and LLaMA-2) with batch sizes of 64, 128, and 256. An active malicious-server threat model was assumed, and analytic gradient inversion was applied to recover text. Reconstruction fidelity was measured over 5 runs using exact sentence accuracy, sentence-level bilingual evaluation understudy (S-BLEU), and recall-oriented understudy for gisting evaluation (ROUGE-L). Results: Exact sentence reconstruction ranged from 27% to 75% across tokenizers, datasets, and batch sizes. At batch size 64 on the discharge dataset, accuracy was 64.7% (GPT-2), 70% (RadBERT), and 67.5% (LLaMA-2), decreasing to 27.3%, 28.5%, and 27.5% at batch size 256. S-BLEU declined with increasing batch size (eg, discharge reports from 0.69 to 0.31). Reconstruction fidelity did not differ significantly across tokenizers (all but one of 27 comparisons nonsignificant; none significant after Holm correction), and approximately 75% of clinical concepts were represented in the reconstructed text (a corpus-level upper bound), regardless of tokenizer. Batch size was the dominant factor governing leakage. Conclusions: Under a worst-case malicious server that tampers with the shared model and observes unprotected per-client gradients (no secure aggregation or differential privacy), substantial portions of radiology-report text can be reconstructed, with up to approximately 75% of reconstructed 32-token sequences and 75% of clinical concepts (not direct patient identifiers) recovered from FL gradients. In a controlled ablation holding model architecture fixed, tokenizer choice, including domain-specific tokenizers, did not significantly affect leakage under the evaluated conditions, whereas batch size was the primary determinant, and no tokenizer significantly reduced the risk. Tokenizer selection should therefore not be treated as a privacy safeguard in this setting. Safeguards such as secure aggregation and differential privacy should therefore be evaluated as candidate protections for FL deployments that must satisfy Health Insurance Portability and Accountability Act (HIPAA) and General Data Protection Regulation (GDPR) requirements in radiology natural language processing (NLP); legal compliance additionally depends on organizational safeguards, risk assessment, and governance beyond the scope of this study.
- 11 September 2026
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 14, e88390
Open AccessPrivacy Leakage in Federated Learning in Radiology Reports: Comparative Evaluation of Tokenizer and Batch-Size Privacy Risks
Background: Federated learning (FL) enables multi-institutional model training on clinical text without sharing raw data; however, gradient inversion methods can reconstruct sensitive information from shared model updates. The extent of such privacy leakage in FL applied to radiology reports, and the role of tokenizer design, remains unclear. Objective: This study aimed to quantify gradient-based reconstruction of radiology report text in an FL setting and to compare privacy risk across 3 transformer tokenization strategies in a controlled, tokenizer-aware evaluation. Methods: Six FL clients trained a GPT-2–style transformer (sequence length 32) on 2 public clinical-text corpora comprising 368,751 diagnostic reports, 98,206 discharge summaries, and 1500 MIMIC-CXR (Medical Information Mart for Intensive Care Chest X-Ray) radiology reports. Models were trained using 3 tokenizers (GPT-2, RadBERT, and LLaMA-2) with batch sizes of 64, 128, and 256. An active malicious-server threat model was assumed, and analytic gradient inversion was applied to recover text. Reconstruction fidelity was measured over 5 runs using exact sentence accuracy, sentence-level bilingual evaluation understudy (S-BLEU), and recall-oriented understudy for gisting evaluation (ROUGE-L). Results: Exact sentence reconstruction ranged from 27% to 75% across tokenizers, datasets, and batch sizes. At batch size 64 on the discharge dataset, accuracy was 64.7% (GPT-2), 70% (RadBERT), and 67.5% (LLaMA-2), decreasing to 27.3%, 28.5%, and 27.5% at batch size 256. S-BLEU declined with increasing batch size (eg, discharge reports from 0.69 to 0.31). Reconstruction fidelity did not differ significantly across tokenizers (all but one of 27 comparisons nonsignificant; none significant after Holm correction), and approximately 75% of clinical concepts were represented in the reconstructed text (a corpus-level upper bound), regardless of tokenizer. Batch size was the dominant factor governing leakage. Conclusions: Under a worst-case malicious server that tampers with the shared model and observes unprotected per-client gradients (no secure aggregation or differential privacy), substantial portions of radiology-report text can be reconstructed, with up to approximately 75% of reconstructed 32-token sequences and 75% of clinical concepts (not direct patient identifiers) recovered from FL gradients. In a controlled ablation holding model architecture fixed, tokenizer choice, including domain-specific tokenizers, did not significantly affect leakage under the evaluated conditions, whereas batch size was the primary determinant, and no tokenizer significantly reduced the risk. Tokenizer selection should therefore not be treated as a privacy safeguard in this setting. Safeguards such as secure aggregation and differential privacy should therefore be evaluated as candidate protections for FL deployments that must satisfy Health Insurance Portability and Accountability Act (HIPAA) and General Data Protection Regulation (GDPR) requirements in radiology natural language processing (NLP); legal compliance additionally depends on organizational safeguards, risk assessment, and governance beyond the scope of this study.
- 24 July 2026
- Published by Springer Nature
- p. 3-33
SoK: A Taxonomy of Attacks and Defenses in Split Learning
- 24 July 2026
- Published by Springer Nature
- p. 3-33
SoK: A Taxonomy of Attacks and Defenses in Split Learning
- Open Access
- 5 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1456-1459
SERSEM: Selective Entropy-Weighted Scoring for Membership Inference in Code Language Models
As Large Language Models (LLMs) for code increasingly utilize massive, often non-permissively licensed datasets, evaluating data contamination through Membership Inference Attacks (MIAs) has become critical. We propose SERSEM (Selective Entropy-Weighted Scoring for Membership Inference), a novel white-box attack framework that suppresses uninformative syntactical boilerplate to amplify specific memorization signals. SERSEM utilizes a dual-signal methodology: first, a continuous character-level weight mask is derived through static Abstract Syntax Tree (AST) analysis, spellchecking-based multilingual logic detection, and offline linting. Second, these heuristic weights are used to pool internal transformer activations and calibrate token-level Z-scores from the output logits. Evaluated on a 25,000-sample balanced dataset, SERSEM achieves a global AUC-ROC of 0.7913 on the StarCoder2-3B model and 0.7867 on the StarCoder2-7B model, consistently outperforming the implemented probability-based baselines Loss, Min-K% Prob, and PAC. Our findings demonstrate that focusing on human-centric coding anomalies provides a significantly more robust indicator of verbatim memorization than sequence-level probability averages.
- 5 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1456-1459
Open AccessSERSEM: Selective Entropy-Weighted Scoring for Membership Inference in Code Language Models
As Large Language Models (LLMs) for code increasingly utilize massive, often non-permissively licensed datasets, evaluating data contamination through Membership Inference Attacks (MIAs) has become critical. We propose SERSEM (Selective Entropy-Weighted Scoring for Membership Inference), a novel white-box attack framework that suppresses uninformative syntactical boilerplate to amplify specific memorization signals. SERSEM utilizes a dual-signal methodology: first, a continuous character-level weight mask is derived through static Abstract Syntax Tree (AST) analysis, spellchecking-based multilingual logic detection, and offline linting. Second, these heuristic weights are used to pool internal transformer activations and calibrate token-level Z-scores from the output logits. Evaluated on a 25,000-sample balanced dataset, SERSEM achieves a global AUC-ROC of 0.7913 on the StarCoder2-3B model and 0.7867 on the StarCoder2-7B model, consistently outperforming the implemented probability-based baselines Loss, Min-K% Prob, and PAC. Our findings demonstrate that focusing on human-centric coding anomalies provides a significantly more robust indicator of verbatim memorization than sequence-level probability averages.
- Open Access
- 29 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Big Data
- Vol. PP (5) , 1-15
Bridging Local and Federated Data Normalization in Federated Learning: A Privacy-Preserving Approach
Data normalization is a crucial preprocessing step for enhancing model performance and training stability. In federated learning (FL), where data remains distributed across multiple parties during collaborative model training, normalization presents unique challenges due to the decentralized and often heterogeneous nature of the data. Traditional methods rely on either independent client-side processing, i.e., local normalization, or normalizing the entire dataset before distributing it to parties, i.e., pooled normalization. Local normalization can be problematic when data distributions across parties are non-IID, while the pooled normalization approach conflicts with the decentralized nature of FL. In this paper, we explore the adaptation of widely used normalization techniques to FL and define the term federated normalization. Federated normalization simulates pooled normalization by enabling the collaborative exchange of normalization parameters among parties. Thus, it achieves performance on par with pooled normalization without compromising data locality. However, sharing normalization parameters such as the mean introduces potential privacy risks, which we further mitigate through a robust privacy-preserving solution. Our contributions include: (i) We systematically evaluate the impact of various federated and local normalization techniques in heterogeneous FL scenarios, (ii) We propose a novel homomorphically encrypted $k$-th ranked element (and median) calculation tailored for the federated setting, enabling secure and efficient federated normalization, (iii) We propose privacy-preserving implementations of widely used normalization techniques for FL, leveraging multiparty fully homomorphic encryption (MHE).
- 29 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Big Data
- Vol. PP (5) , 1-15
Open AccessBridging Local and Federated Data Normalization in Federated Learning: A Privacy-Preserving Approach
Data normalization is a crucial preprocessing step for enhancing model performance and training stability. In federated learning (FL), where data remains distributed across multiple parties during collaborative model training, normalization presents unique challenges due to the decentralized and often heterogeneous nature of the data. Traditional methods rely on either independent client-side processing, i.e., local normalization, or normalizing the entire dataset before distributing it to parties, i.e., pooled normalization. Local normalization can be problematic when data distributions across parties are non-IID, while the pooled normalization approach conflicts with the decentralized nature of FL. In this paper, we explore the adaptation of widely used normalization techniques to FL and define the term federated normalization. Federated normalization simulates pooled normalization by enabling the collaborative exchange of normalization parameters among parties. Thus, it achieves performance on par with pooled normalization without compromising data locality. However, sharing normalization parameters such as the mean introduces potential privacy risks, which we further mitigate through a robust privacy-preserving solution. Our contributions include: (i) We systematically evaluate the impact of various federated and local normalization techniques in heterogeneous FL scenarios, (ii) We propose a novel homomorphically encrypted $k$-th ranked element (and median) calculation tailored for the federated setting, enabling secure and efficient federated normalization, (iii) We propose privacy-preserving implementations of widely used normalization techniques for FL, leveraging multiparty fully homomorphic encryption (MHE).
- Open Access
- 24 November 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Privacy Leakage in Federated Learning in Radiology Reports: A Comparative Evaluation of Tokenizer-Driven Privacy Risks (Preprint)
BACKGROUND Federated learning (FL) enables multi-institutional model training on clinical text without sharing raw data; however, gradient inversion methods can reconstruct sensitive information from shared model updates. The extent of such privacy leakage in FL applied to radiology reports, and the role of tokenizer design, remains unclear. OBJECTIVE To quantify gradient-based reconstruction of radiology report text in an FL setting and to compare privacy risk across three transformer tokenization strategies in a controlled, tokenizer-aware evaluation. METHODS Six FL clients trained a GPT-2–style transformer (117M parameters; sequence length 32) on two public radiology corpora comprising 368,751 diagnostic reports, 98,206 discharge summaries, and 1,500 MIMIC-CXR free-text reports. Models were trained using three tokenizers (GPT-2, RadBERT, LLaMA-2) with batch sizes of 64, 128, and 256. A curious-server threat model was assumed, and analytic gradient inversion was applied to recover text. Reconstruction fidelity was measured over five runs using exact sentence accuracy, S-BLEU, and ROUGE-L. RESULTS Exact sentence reconstruction ranged from 33% to 42% across tokenizers. At batch size 64, accuracy was 42.1% (GPT-2), 42.3% (RadBERT), and 39.4% (LLaMA-2), decreasing to 37.3%, 37.2%, and 34.3% at batch size 256. S-BLEU scores declined with increasing batch size (e.g., GPT-2: 0.44→0.33; RadBERT: 0.48→0.35; LLaMA-2: 0.39→0.30). RadBERT yielded higher reconstruction fidelity and greater recovery of clinical terms, but no tokenizer prevented leakage. CONCLUSIONS Substantial portions of radiology report text can be reconstructed from FL gradients even with larger batch sizes and domain-specific tokenizers. Tokenizer design influences leakage severity and should be incorporated into privacy evaluations for clinical language models. Integrating safeguards such as secure aggregation and differential privacy is necessary to meet HIPAA and GDPR requirements when deploying FL for radiology NLP. CLINICALTRIAL Not applicable.
- 24 November 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessPrivacy Leakage in Federated Learning in Radiology Reports: A Comparative Evaluation of Tokenizer-Driven Privacy Risks (Preprint)
BACKGROUND Federated learning (FL) enables multi-institutional model training on clinical text without sharing raw data; however, gradient inversion methods can reconstruct sensitive information from shared model updates. The extent of such privacy leakage in FL applied to radiology reports, and the role of tokenizer design, remains unclear. OBJECTIVE To quantify gradient-based reconstruction of radiology report text in an FL setting and to compare privacy risk across three transformer tokenization strategies in a controlled, tokenizer-aware evaluation. METHODS Six FL clients trained a GPT-2–style transformer (117M parameters; sequence length 32) on two public radiology corpora comprising 368,751 diagnostic reports, 98,206 discharge summaries, and 1,500 MIMIC-CXR free-text reports. Models were trained using three tokenizers (GPT-2, RadBERT, LLaMA-2) with batch sizes of 64, 128, and 256. A curious-server threat model was assumed, and analytic gradient inversion was applied to recover text. Reconstruction fidelity was measured over five runs using exact sentence accuracy, S-BLEU, and ROUGE-L. RESULTS Exact sentence reconstruction ranged from 33% to 42% across tokenizers. At batch size 64, accuracy was 42.1% (GPT-2), 42.3% (RadBERT), and 39.4% (LLaMA-2), decreasing to 37.3%, 37.2%, and 34.3% at batch size 256. S-BLEU scores declined with increasing batch size (e.g., GPT-2: 0.44→0.33; RadBERT: 0.48→0.35; LLaMA-2: 0.39→0.30). RadBERT yielded higher reconstruction fidelity and greater recovery of clinical terms, but no tokenizer prevented leakage. CONCLUSIONS Substantial portions of radiology report text can be reconstructed from FL gradients even with larger batch sizes and domain-specific tokenizers. Tokenizer design influences leakage severity and should be incorporated into privacy evaluations for clinical language models. Integrating safeguards such as secure aggregation and differential privacy is necessary to meet HIPAA and GDPR requirements when deploying FL for radiology NLP. CLINICALTRIAL Not applicable.
- Open Access
- 24 November 2025
- Published by Springer Nature in Genome Biology
- Vol. 26 (1) , 400
A reinforcement learning-based approach for dynamic privacy protection in genomic data sharing beacons
The rise of genomic sequencing raises privacy concerns due to the identifiable nature of genomic data. The GA4GH Beacon Project enables privacy-preserving data sharing but is vulnerable to membership inference attacks that reveal individual participation. Existing defenses, such as noise addition and query restrictions, rely on static policies that attackers can bypass. We introduce the first reinforcement learning (RL)-based dynamic defense for the beacon protocol, training defender and attacker agents in a multiplayer setting. Our approach adapts responses in real time, distinguishing users from adversaries and balancing privacy with utility against evolving threats.
- 24 November 2025
- Published by Springer Nature in Genome Biology
- Vol. 26 (1) , 400
Open AccessA reinforcement learning-based approach for dynamic privacy protection in genomic data sharing beacons
The rise of genomic sequencing raises privacy concerns due to the identifiable nature of genomic data. The GA4GH Beacon Project enables privacy-preserving data sharing but is vulnerable to membership inference attacks that reveal individual participation. Existing defenses, such as noise addition and query restrictions, rely on static policies that attackers can bypass. We introduce the first reinforcement learning (RL)-based dynamic defense for the beacon protocol, training defender and attacker agents in a multiplayer setting. Our approach adapts responses in real time, distinguishing users from adversaries and balancing privacy with utility against evolving threats.
- Open Access
Bridging Local and Federated Data Normalization in Federated Learning: A Privacy-Preserving Approach
Data normalization is a crucial preprocessing step for enhancing model performance and training stability. In federated learning (FL), where data remains distributed across multiple parties during collaborative model training, normalization presents unique challenges due to the decentralized and often heterogeneous nature of the data. Traditional methods rely on either independent client-side processing, i.e., local normalization, or normalizing the entire dataset before distributing it to parties, i.e., pooled normalization. Local normalization can be problematic when data distributions across parties are non-IID, while the pooled normalization approach conflicts with the decentralized nature of FL. In this paper, we explore the adaptation of widely used normalization techniques to FL and define the term federated normalization. Federated normalization simulates pooled normalization by enabling the collaborative exchange of normalization parameters among parties. Thus, it achieves performance on par with pooled normalization without compromising data locality. However, sharing normalization parameters such as the mean introduces potential privacy risks, which we further mitigate through a robust privacy-preserving solution. Our contributions include: (i) We systematically evaluate the impact of various federated and local normalization techniques in heterogeneous FL scenarios, (ii) We propose a novel homomorphically encrypted $k$-th ranked element (and median) calculation tailored for the federated setting, enabling secure and efficient federated normalization, (iii) We propose privacy-preserving implementations of widely used normalization techniques for FL, leveraging multiparty fully homomorphic encryption (MHE).
Open AccessBridging Local and Federated Data Normalization in Federated Learning: A Privacy-Preserving Approach
Data normalization is a crucial preprocessing step for enhancing model performance and training stability. In federated learning (FL), where data remains distributed across multiple parties during collaborative model training, normalization presents unique challenges due to the decentralized and often heterogeneous nature of the data. Traditional methods rely on either independent client-side processing, i.e., local normalization, or normalizing the entire dataset before distributing it to parties, i.e., pooled normalization. Local normalization can be problematic when data distributions across parties are non-IID, while the pooled normalization approach conflicts with the decentralized nature of FL. In this paper, we explore the adaptation of widely used normalization techniques to FL and define the term federated normalization. Federated normalization simulates pooled normalization by enabling the collaborative exchange of normalization parameters among parties. Thus, it achieves performance on par with pooled normalization without compromising data locality. However, sharing normalization parameters such as the mean introduces potential privacy risks, which we further mitigate through a robust privacy-preserving solution. Our contributions include: (i) We systematically evaluate the impact of various federated and local normalization techniques in heterogeneous FL scenarios, (ii) We propose a novel homomorphically encrypted $k$-th ranked element (and median) calculation tailored for the federated setting, enabling secure and efficient federated normalization, (iii) We propose privacy-preserving implementations of widely used normalization techniques for FL, leveraging multiparty fully homomorphic encryption (MHE).
- Open Access
- 19 May 2025
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 41 (6)
Beacon Reconstruction Attack: Reconstruction of genomes in genomic data-sharing beacons using summary statistics
Motivation: Genomic data-sharing beacon protocol, developed by the Global Alliance for Genomics and Health, offers a privacy-preserving mechanism for querying genomic datasets while restricting direct data access. Despite their design, beacons remain vulnerable to privacy attacks. This study introduces a novel privacy vulnerability of the protocol: one can reconstruct large portions of the genomes of all beacon participants by only using the summary statistics reported by the protocol. Results: We introduce a novel optimization-based algorithm that leverages beacon responses and SNP correlations for reconstruction. By optimizing for the SNP correlations and allele frequencies, the proposed approach achieves genome reconstruction with a substantially higher F1-score (70%) compared to baseline methods (45%) on beacons generated using individuals from the HapMap and OpenSNP datasets. We show that reconstructed genomes can be used by downstream applications such as in membership inference attacks against other beacons. Our findings reveal that beacons releasing allele frequencies substantially increase the reconstruction risk, underscoring the need for enhanced privacy-preserving mechanisms to protect genomic data. Availability and implementation: Our implementation is available at https://github.com/ASAP-Bilkent/Beacon-Reconstruction-Attack.
- 19 May 2025
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 41 (6)
Open AccessBeacon Reconstruction Attack: Reconstruction of genomes in genomic data-sharing beacons using summary statistics
Motivation: Genomic data-sharing beacon protocol, developed by the Global Alliance for Genomics and Health, offers a privacy-preserving mechanism for querying genomic datasets while restricting direct data access. Despite their design, beacons remain vulnerable to privacy attacks. This study introduces a novel privacy vulnerability of the protocol: one can reconstruct large portions of the genomes of all beacon participants by only using the summary statistics reported by the protocol. Results: We introduce a novel optimization-based algorithm that leverages beacon responses and SNP correlations for reconstruction. By optimizing for the SNP correlations and allele frequencies, the proposed approach achieves genome reconstruction with a substantially higher F1-score (70%) compared to baseline methods (45%) on beacons generated using individuals from the HapMap and OpenSNP datasets. We show that reconstructed genomes can be used by downstream applications such as in membership inference attacks against other beacons. Our findings reveal that beacons releasing allele frequencies substantially increase the reconstruction risk, underscoring the need for enhanced privacy-preserving mechanisms to protect genomic data. Availability and implementation: Our implementation is available at https://github.com/ASAP-Bilkent/Beacon-Reconstruction-Attack.
- Open Access
A Taxonomy of Attacks and Defenses in Split Learning
Split Learning (SL) has emerged as a promising paradigm for distributed deep learning, allowing resource-constrained clients to offload portions of their model computation to servers while maintaining collaborative learning. However, recent research has demonstrated that SL remains vulnerable to a range of privacy and security threats, including information leakage, model inversion, and adversarial attacks. While various defense mechanisms have been proposed, a systematic understanding of the attack landscape and corresponding countermeasures is still lacking. In this study, we present a comprehensive taxonomy of attacks and defenses in SL, categorizing them along three key dimensions: employed strategies, constraints, and effectiveness. Furthermore, we identify key open challenges and research gaps in SL based on our systematization, highlighting potential future directions.
Open AccessA Taxonomy of Attacks and Defenses in Split Learning
Split Learning (SL) has emerged as a promising paradigm for distributed deep learning, allowing resource-constrained clients to offload portions of their model computation to servers while maintaining collaborative learning. However, recent research has demonstrated that SL remains vulnerable to a range of privacy and security threats, including information leakage, model inversion, and adversarial attacks. While various defense mechanisms have been proposed, a systematic understanding of the attack landscape and corresponding countermeasures is still lacking. In this study, we present a comprehensive taxonomy of attacks and defenses in SL, categorizing them along three key dimensions: employed strategies, constraints, and effectiveness. Furthermore, we identify key open challenges and research gaps in SL based on our systematization, highlighting potential future directions.
- 9 February 2025
- Published by Springer Nature
- p. 235-244
Client Security Alone Fails in Federated Learning: 2D and 3D Attack Insights
- 9 February 2025
- Published by Springer Nature
- p. 235-244
Client Security Alone Fails in Federated Learning: 2D and 3D Attack Insights