--- title: Neuron Boundary Coverage (NBC) in DNN Testing url: https://www.emergentmind.com/topics/neuron-boundary-coverage-nbc type: topic --- # Neuron Boundary Coverage (NBC) in DNN Testing Neuron Boundary Coverage (NBC) is a structural test coverage metric designed for white-box evaluation of deep neural networks (DNNs). Its primary objective is to quantify how thoroughly a test suite explores activation behaviors that were not encountered during training, by measuring the frequency that test inputs drive a neuron’s activation outside its empirically observed boundaries. NBC is used to identify coverage of "corner-case" activations, offering insights into model robustness, safety, and exposure to novel or potentially adversarial behaviors [2208.03407][2505.08814]. ## 1. Formal Definition and Coverage Computation For a trained neural network $N$ with $L$ layers, let $a_{l,i}(x)$ denote the activation of neuron $i$ in layer $l$ on input $x$. During a profiling stage, a dataset $\mathcal{T}_{\mathrm{prof}}$ (typically the training set) is used to record observed minimum and maximum activations for each neuron: - $low_{l,i} = \min_{x \in \mathcal{T}_{\mathrm{prof}}} a_{l,i}(x)$ - $high_{l,i} = \max_{x \in \mathcal{T}_{\mathrm{prof}}} a_{l,i}(x)$ Each neuron $(l,i)$ thus acquires a "main functional region" $[low_{l,i}, high_{l,i}]$, representing typical activation values during training. NBC coverage obligations are incurred whenever a test input $x$ pushes a neuron’s activation below $low_{l,i}$ or above $high_{l,i}$. Define the set of NBC test conditions as: $$ C_{NBC} = \{ (l,i,\mathrm{dir}) \mid 1 high_{l,i}$, or - $\mathrm{dir} = \text{``below''}$ and $a_{l,i}(x) < low_{l,i}$. The NBC coverage ratio is: $$ NBC(T) = \frac{\Bigl|\{(l,i,{\rm dir})\in C_{NBC}\;\mid\;\exists x\in T:\; ({\rm dir}=\text{``above''}\wedge a_{l,i}(x)>high_{l,i})\vee ({\rm dir}=\text{``below''}\wedge a_{l,i}(x) high_{l,i} + \epsilon$ Varying $\epsilon$ provides a family of stricter or looser coverage criteria. Canonical NBC uses $\epsilon=0$ [2505.08814]. ### Example Calculation Consider a network with one hidden layer of two neurons: | Neuron | $low$ | $high$ | |-----------|---------|---------| | (2,1) | 0.0 | 1.0 | | (2,2) | –0.5 | 0.5 | Test suite $T = \{x_1, x_2, x_3\}$ yields activations: | Input | $a_{2,1}(x)$ | $a_{2,2}(x)$ | |-------|--------------|--------------| | $x_1$ | 0.4 | 0.0 | | $x_2$ | 1.2 | 0.1 | | $x_3$ | 0.8 | –0.7 | Obligations covered: - (2,1,above): $x_2$ (1.2>1.0) - (2,2,below): $x_3$ (–0.7<–0.5) Two of four obligations are met, thus $NBC(T) = 50\%$ [2208.03407]. ## 3. Empirical Characteristics and Model Sensitivity Empirical studies underscore that NBC values remain very low on standard test data. For instance, NBC for LeNet-1 on MNIST is 0.87%, and for ResNet20 on CIFAR-10, 5.55% [2208.03407]. Expanded experiments investigating networks of depths 5–54 (LeNet, VGG, ResNet) display the following: - At $\epsilon=0$, NBC is under 1% for most models and decreases with increasing network depth in purely sequential architectures. - For negative $\epsilon$ (requiring activations strictly inside the training region), NBC increases but quickly falls as $\epsilon \to 0$. - In residual architectures (deep ResNets), shallow-margin NBC may be slightly higher, plausibly due to more extreme-activation neurons in very deep architectures [2505.08814]. The coverage obtained also depends on the architecture and dataset: - LeNet's NBC decays more gradually with $\epsilon$ than VGG or ResNet on CIFAR-10. - Upper-boundary (SNAC) coverage tends to substantially exceed lower-boundary coverage, especially in CIFAR-10-trained models, indicating a higher occurrence of above-max test-time activations [2505.08814]. ## 4. Comparative Positioning Among Structural Metrics NBC is stricter than upper-bound-only metrics (e.g., SNAC), which consider only the event of a neuron exceeding its observed maximum. However, NBC is weaker in granularity than k-multisection neuron coverage (KMNC): while NBC focuses exclusively on out-of-bounds excursions, KMNC subdivides the main activation region and tracks coverage of each subregion, resulting in higher and more informative scores (e.g., $KMNC@10$ over 90%, NBC $\ll$ 10%) [2208.03407][2505.08814]. Variants and extensions include: - SNAC: Strong Neuron Activation Coverage, measuring upper-boundary coverage only. - Margin-augmented NBC: Employing $\epsilon > 0$ yields a stricter (lower) coverage curve; as $\epsilon$ increases, coverage quickly drops to near zero. NBC, SNAC, and KMNC are often jointly reported to establish a fuller and more nuanced assessment of a test suite's thoroughness. ## 5. Practical Issues and Limitations Several practical considerations affect NBC's utility and interpretation: - **Boundary estimation** is critically dependent on the representativeness of $\mathcal{T}_{\mathrm{prof}}$; unrepresentative profiling can yield excessively loose or tight boundaries and hence misleading coverage. - **Activation function sensitivity** means that metrics behave differently by architecture. ReLU activations (where $low=0$) rarely yield lower-boundary hits, except for numerical effects or negative-leak variants. Sigmoid/tanh activations yield finite $[low, high]$ intervals for both sides enabling more sensitive NBC measurement. - **Layer and neuron selection** scales the number of obligations as twice the total neuron count over nominated layers; limiting NBC to certain layers or neuron subsets may be preferred for tractability. NBC alone is too coarse for coverage-guided test generation, as its absolute values are low and it may not provide fine-grained guidance, but it highlights the risk of untested neuron activation regimes [2208.03407][2505.08814]. ## 6. Tool Support and Measurement in DNNCov NBC is implemented in the DNNCov tool [2208.03407]: - **Profiling phase:** Passes the training set through the network, logs per-neuron minima and maxima. - **Coverage evaluation:** For each test input and neuron, checks for out-of-boundary activations and records covered obligations. - **Reporting:** Computes $NBC(T)$ and provides visualization of out-of-bounds neuron hits, with quantitative min/avg/max reports per obligation. Example DNNCov usage: ``` python dnncov.py --model mynet.h5 --criteria nbc --mnist-dataset --train 60000 --test 10000 --outputs outs_nbc ``` After processing, the analysis specifies which test inputs triggered out-of-boundary activations, supporting researchers’ auditing and test suite refinement [2208.03407]. ## 7. Interpretations, Patterns, and Future Directions Empirical results suggest that neurons are rarely driven outside their training activation range by standard test suites, and model depth generally reduces NBC in standard architectures. SNAC systematically yields higher values than NBC, reflecting a consistent asymmetry: above-boundary activations are more prevalent than below-boundary, especially in deep architectures and on challenging datasets. Directions for future research, as highlighted, include: - Theorizing neuron "main functional" and "boundary" regions to offer formal robustness guarantees, - Investigating synergistic combinations of coverage metrics for guided test generation, - Extending NBC to deep, non-sequential, or non-feedforward architectures (e.g., transformers, graph networks) by adopting distribution-aware or layer-wise intervals [2505.08814]. NBC serves as a direct, empirically grounded measure to probe whether new or adversarially generated test cases exercise activations beyond the network’s prior experience, strengthening structural coverage analysis when used in concert with other metrics. Source: https://www.emergentmind.com/topics/neuron-boundary-coverage-nbc