- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 1
Read review
Paper summary
This paper observes that benign models typically exhibit consistent channel activation magnitudes across samples, whereas malicious clients exhibit substantial differences in channel activation magnitudes between clean and malicious samples. Some channels respond strongly to malicious samples but weakly to clean samples, while others exhibit the opposite pattern. Based on this observation, the paper proposes CAS. During local training, clients compute and upload the mean channel activations for each class. The server constructs channel scores using TF-IDF representations and cross-class entropy, then clusters clients using the Wasserstein distance. The cluster whose centroid has the larger norm is retained for aggregation. Experiments on four datasets under non-IID data distributions demonstrate that the method provides effective defense against four different attacks compared with baseline defenses, while incurring a relatively small loss in accuracy.
Critical strengths
1.The paper presents a novel perspective on federated backdoor defense. Instead of relying on gradient or parameter discrepancies, CAS detects malicious clients by measuring abnormal class channel activation patterns using TF-IDF and activation entropy.
2.The experimental evaluation cover multiple datasets, backdoor attacks, defense, architectures, and federated learning settings.
Critical weaknesses
1.CAS requires clients to upload both model updates and class-wise mean channel activations. However, under the paper’s threat model, malicious clients that can manipulate model updates should also be able to manipulate the accompanying activation statistics. An attacker could use local clean data or a separately trained clean model to generate benign-looking CAS statistics while uploading a backdoored model inconsistent with those statistics, thereby bypassing server-side clustering. The BatchNorm mean consistency check proposed in the appendix also appears insufficient to address this issue, since both the activation statistics and the BatchNorm running means are controlled by the client and can be jointly fabricated.
2.I have concerns about the conclusion regarding channel activation inconsistency supported by Figure 2. Figure 2(d) compares two sets of clean data on a benign model, whereas Figures 2(a)–(c) compare clean and triggered data on backdoored models. Thus, both the model state and the input conditions change simultaneously. The current results cannot distinguish whether the observed activation differences arise from the backdoor embedded in the model or merely from the presence of the trigger in the input.
3.The paper does not explain how CAS handles missing classes under highly non-IID data distributions. Because CAS requires class-wise channel activation statistics, clients without samples from certain classes may naturally exhibit concentrated, low-entropy activation patterns and consequently be misclassified as malicious.
Questions for authors
see weaknesses
7 comments
Response to Reviewer oWNB (2/3)
Response to Critical Weakness 2:
- "I have concerns about the conclusion regarding channel activation inconsistency supported by Figure 2. Figure 2(d) compares two sets of clean data on a benign model, whereas Figures 2(a)–(c) compare clean and triggered data on backdoored models. Thus, both the model state and the input conditions change simultaneously."
We appreciate that the reviewer asks this question. We agree with the reviewer that the observation of channel activation inconsistency supported by Figure 2 comes from two different type of data consisting between the backdoor and benign models. To enhance our claim, we have replaced the Figure 2 (d) to another one (link) that instead shows the activation of a benign model using poisoned . In the revised manuscript, we have updated the setup in Section 2.2 line 152 to replace the figure 2 (d) in section 2.2 line 162.
- "The current results cannot distinguish whether the observed activation differences arise from the backdoor embedded in the model or merely from the presence of the trigger in the input"
We agree with the reviewer that the results in Figure 2 can cause ambiguity in the resulted activation differences. To address this problem, we further investigate the scenario where the malicious client submit the CAS statistics computed on clean (no trigger injected) using the backdoored model. The resulted plot (link) reveals another key finding: The benign model’s channels form a compact cluster, with every channel above about 0.6. In contrast, the backdoored model’s activations are spread over a much wider range (about 0.2–1.7), and a large number of channels falls below the lowest activation of any channel in the benign model. These finding reveals that, although clean samples do not active the channel with backdoor embedded, they rather remain dormant on clean samples by showing a wider spread and lower activation magnitude. Therefore, we can conclude that backdoor embedded in the model can always result in activation differences compared benign models no matter to the presence of the trigger in the input. In this revised manuscript, we have created a new section to included the additional results and explanation to our channel activation inconsistency in Appendix B line 842.
Moreover, even under the scenario where malicious clients submit the CAS statistics computed purely on clean samples, the TF-IDF can still effectively capture the abnormal pattern from them. Since the activations have a wider spread across different channels, the TF value will also be spread, resulting abnormal CAS score distribution compared to benign ones. The following table shows the results when malicious clients submit the CAS statistics computed clean samples:
Table R3-2: Performance of CAS when malicious clients compute CAS statistics on clean samples
| Dataset | Attack | Acc | ASR | TPR | FPR |
|---|---|---|---|---|---|
| CIFAR-10 | BadPFL | 95.13 ± 0.32 | 13.02 ± 2.87 | 93.30 ± 1.92 | 8.03 ± 3.09 |
| CIFAR-10 | IBA | 95.12 ± 0.32 | 10.16 ± 1.00 | 97.40 ± 1.85 | 4.83 ± 2.05 |
| CIFAR-10 | DBA | 95.07 ± 0.33 | 10.05 ± 0.13 | 99.40 ± 1.08 | 3.92 ± 1.86 |
| CIFAR-10 | Neurotoxin | 95.14 ± 0.34 | 10.02 ± 0.13 | 99.20 ± 0.45 | 4.30 ± 1.07 |
| CIFAR-100 | BadPFL | 78.56 ± 0.81 | 1.97 ± 1.16 | 95.40 ± 1.71 | 1.30 ± 1.06 |
| CIFAR-100 | IBA | 78.49 ± 0.59 | 1.23 ± 0.19 | 99.70 ± 0.45 | 0.13 ± 0.15 |
| CIFAR-100 | DBA | 78.38 ± 0.76 | 1.11 ± 0.03 | 99.80 ± 0.27 | 0.18 ± 0.11 |
| CIFAR-100 | Neurotoxin | 78.39 ± 0.73 | 1.04 ± 0.03 | 99.80 ± 0.27 | 0.93 ± 0.39 |
As shown in Table R1-3, CAS stays effective under this adaptive attack. On both datasets it detects at least 93% of malicious clients and keeps ASR close to random guess. This shows that CAS does not depend on whether the class-wise activation statistics are computed on triggered or clean samples. The backdoor embedded in the model alone is enough to produce detectable anomalies in the CAS scores. In the revised manuscript, we have incorporated this adaptive attacker scenario in section 3.1 line 467.
Response to Reviewer oWNB (3/3)
Response to Critical Weakness 3:
We thank the reviewer for this comment. As demonstrated above, benign clients' channels are lower bounded given the premise that they capture features from the same input distribution. However, channels with backdoor embedded are inevitably deviated from the normal channel by showing abnormal dormancy, resulting lower activation entropy regardless to data distribution. In this case, the cluster grouped by malicious clients will often have a lower center norm due to their lower activation entropy. Moreover, since malicious clients often have fixed malicious objective (e.g., same target class), they inevitably tend to have aligned lower entropy behaviors towards the target class. In contrast, the low entropy behaviors of all benign clients often do not bind to a particular class. In such case, our proposed clustering mechanism can effectively capture the CAS abnormal pattern among all malicious clients and exclude them from aggregation. To support our claim, we provide the experimental results showing the superior defense performance of CAS under various degree of data heterogeneity:
Table R3-3: Performance of CAS under different degrees of data heterogeneity (Dirichlet ) on CIFAR-100
| Attack | ACC | ASR ↓ | TPR ↑ | FPR ↓ | |
|---|---|---|---|---|---|
| 0.1 | DBA | 75.43 | 0.94 | 100.00 | 0.63 |
| 0.1 | IBA | 75.72 | 0.82 | 100.00 | 0.00 |
| 0.1 | BadPFL | 76.01 | 1.14 | 96.50 | 1.00 |
| 0.1 | Neurotoxin | 75.70 | 0.96 | 100.00 | 0.37 |
| 0.3 | DBA | 76.62 | 1.14 | 100.00 | 0.13 |
| 0.3 | IBA | 77.41 | 1.03 | 98.00 | 0.25 |
| 0.3 | BadPFL | 77.35 | 1.66 | 93.50 | 1.38 |
| 0.3 | Neurotoxin | 76.87 | 1.07 | 100.00 | 0.13 |
| 0.5 | DBA | 77.76 | 1.14 | 100.00 | 0.00 |
| 0.5 | IBA | 77.62 | 0.95 | 99.50 | 0.13 |
| 0.5 | BadPFL | 77.92 | 1.33 | 97.50 | 0.13 |
| 0.5 | Neurotoxin | 77.71 | 1.13 | 100.00 | 0.25 |
Experiment logs for Table R3-1 to R3-3 can be found in here.
Once again, we sincerely thank Reviewer oWNB for his/her time and the constructive and insightful comments. We look forward to any further discussion with you.
References:
[1] Li et al., "BackdoorIndicator: Leveraging OOD data for proactive backdoor detection in federated learning", USENIX Security 2024.
[2] Kadir et al., "Backdoor Attacks Rerouted: Batch Normalization as a Sink for Adversarial Signals", NeurIPS 2026.
[3] Huang et al., "Lockdown: Backdoor Defense for Federated Learning with Isolated Subspace Training", NeurIPS 2023.
[4] Huang et al., "Parameter Disparities Dissection for Backdoor Defense in Heterogeneous Federated Learning", NeurIPS 2024.
[5] Yan et al., "FedRACE: A Hierarchical and Statistical Framework for Robust Federated Learning", NeurIPS 2025
Response to Reviewer oWNB (1/3)
We thank
reviewer oWNBfor the feedback and valuable suggestions. Below we provide responses to each weakness and question. Please feel free to reach out if further clarification is needed.Response to Critical Weakness 1:
We thank the reviewer for pointing this out. We agree with the reviewer that an adaptive attacker can bypass the BN statistics authentication by simultaneously manipulating the BN layers. However, as suggested in [1] [2], BN layers often encode a substantial amount of backdoor information to enable effective triggered behavior. That is, the training data of a backdoor task can be equivalently seen as a form of out-of-distribution data relative to that of the main task. Due to change of the training data, the estimated mean μ and variance σ after the backdoor task gradually deviate from those of clean task. Thus, in order to bypass the BN authentication, attackers must confine to upload the BN statistics that are not longer capable to maintain the backdoor effect. In other words, the proposed BN authentication aims to drive the adaptive attackers into a dilemma where they either get caught by the authentication or suffer from severe backdoor performance degradation. To support our claim, we consider three forms of adaptive attackers:
Table R3-1: Performance of BN authentication against adaptive attackers in CIFAR10
As shown, CAS with BN authentication enabled provides effective defense service under all adaptive attack scenarios, minimizing the ASR rate down to random guess level. Other FL defenses that additionally requires local clients to upload defensive statistics (e.g., Lockdown[3], FDCR[4], FEDRACE[5]) often do not offer a practice solution for the aforementioned adaptive attackers or simply assuming that the local gradients are unbreakable. In contrast, CAS naturally provides the guaranteed robustness under adaptive attackers. In the revised manuscript, we have incorporated aforementioned justification and results in Appendix F line 1066.