--- title: Planted Anomalous Community url: https://www.emergentmind.com/topics/planted-anomalous-community type: topic --- # Planted Anomalous Community A planted anomalous community is a hidden subset of vertices whose induced interactions deviate systematically from a background random graph or hypergraph model. In the canonical sparse-graph formulation, one observes \(G \sim \mathcal{G}(N,q)\) under the null and \(G \sim \mathcal{G}(N,K,p,q)\) under the alternative, where a hidden subset \(S \subset [N]\) of mean size \(K\) has within-community edge probability \(p=q_{\mathrm{in}}=cq\) with \(c>1\), while all other edges occur with probability \(q=N^{-\alpha}\); with \(K=\Theta(N^\beta)\), the problem exhibits a sharp computational-statistical theory governed by \(\alpha\) and \(\beta\) [1406.6625]. Closely related formulations replace Bernoulli edges by Gaussian weights, hyperedges, directed comparisons, or circular phases, but retain the same organizing idea: a small latent subset induces a coherent deviation relative to a null ensemble [2011.11500] [2411.19885] [2601.05993]. ## 1. Formal definition and canonical model In the Erdős–Rényi background model, \(G \sim \mathcal{G}(N,q)\) has independent edges, each present with probability \(q\). The planted anomalous community model introduces a hidden subset \(S \subset [N]\), \(|S|=K\), such that edges within \(S\) appear with elevated probability \(p=cq\), while all other edges remain at probability \(q\). The null and alternative are \[ H_0: G \sim \mathcal{G}(N,q), \qquad H_1: G \sim \mathcal{G}(N,K,p,q), \] and the paper’s main regime is \(p=cq=\Theta(N^{-\alpha})\), \(K=\Theta(N^\beta)\), with \(c>1\) fixed, \(\alpha \in [0,2]\), and \(\beta \in [0,1]\). The planted subgraph size is random in the main formulation: each vertex is included in \(S\) independently with probability \(K/N\), so \(|S| \sim \mathrm{Binomial}(N,K/N)\) with mean \(K\); this random-size model is statistically equivalent, up to constants, to the fixed-size model in the regimes considered. The basic signal quantities are \[ \mathbb{E}[E(S)] = \binom{K}{2}p,\qquad \mathbb{E}_0[E(S)] = \binom{K}{2}q,\qquad \Delta_S = \binom{K}{2}(p-q)=\binom{K}{2}(c-1)q. \] Increasing \(\alpha\) or decreasing \(\beta\) makes detection harder [1406.6625]. Within this literature, “planted anomalous community” and “planted dense subgraph” are effectively synonymous in the graph case. The same structural role appears in the weighted \(d\)-uniform planted \(k\)-densest sub-hypergraph model, where the planted set induces a coherent mean shift \(\beta\) on the \(\binom{k}{d}\) hyperedges entirely within the community, and the maximum-likelihood estimator is the \(k\)-densest sub-hypergraph objective \(\arg\max_{\hat x \in \mathcal{C}_{p,k}} \langle Y,\hat x^{\otimes d}\rangle\) [2011.11500]. In directed graphs, the anomaly can instead be a ranked community: the planted subset is not distinguished by higher edge density, but by unusual consistency of pairwise orderings [2411.19885]. In circular-data models, the planted community is defined by phase coherence on the internal edges of a complete graph, with signal edges drawn either from a common arc of length \(2\pi\tau\) or from a von Mises distribution with common location parameter \(\mu\) [2601.05993]. ## 2. Statistical detectability in sparse Erdős–Rényi graphs For the Bernoulli planted dense subgraph model with \(p=cq\), the sharp information-theoretic detection threshold is \[ \beta^{\ast}(\alpha)=\alpha \wedge \left(\frac{1}{2}+\frac{\alpha}{4}\right). \] If \(\beta < \beta^{\ast}(\alpha)\), no test of any computational complexity can reliably distinguish \(H_1\) from \(H_0\); the total variation distance tends to zero. If \(\beta > \beta^{\ast}(\alpha)\), there exists a generally computationally intensive test achieving vanishing error probability. A non-asymptotic impossibility statement is given by Proposition 1: if \[ q \le C\left(\frac{1}{K}\log\frac{eN}{K}\ \wedge\ \frac{N^2}{K^4}\right), \] then \[ d_{\mathrm{TV}}(P_0,P_1) \le h(Cc^2)+e^{-K/8}, \] for some function \(h\) with \(h(0+)=0\), so detection is impossible in that regime [1406.6625]. The same result is often summarized through three regions in the \((\alpha,\beta)\)-plane. The “simple regime” is \[ \beta>\frac{1}{2}+\frac{\alpha}{4}, \] where a linear-time test based on total edge count is statistically optimal. The “hard regime” is \[ \alpha<\beta<\frac{1}{2}+\frac{\alpha}{4}, \] where detection is statistically possible via scan statistics or similar exhaustive procedures but is conjectured to be computationally intractable. The “impossible regime” is \[ \beta<\min\left\{\alpha,\frac{1}{2}+\frac{\alpha}{4}\right\}, \] where detection is information-theoretically impossible. The critical sparsity \[ \alpha^\star=\frac{2}{3} \] marks the point where the information-theoretic boundary and the efficient frontier coincide: when \(\alpha \ge 2/3\), \(\beta^\ast(\alpha)=1/2+\alpha/4\), and no computational-statistical gap remains; when \(\alpha<2/3\), the statistical boundary is \(\beta^\ast(\alpha)=\alpha\), which is strictly below the efficient boundary [1406.6625]. This phase diagram is specific to the constant-factor anomaly \(p=cq\), but the organizing principle is broader. A plausible implication is that sparsity does not simply make the problem uniformly harder; rather, it changes which statistic is optimal and whether exhaustive combinatorial structure is necessary. ## 3. Efficient procedures, scan methods, and computational hardness Two explicit test statistics organize the graph case. The linear-time procedure uses the total number of edges, \[ T_{\mathrm{lin}}=\sum_{i\tau_1\), runs in \(O(N+|E|)\), and succeeds when \[ \frac{K^4q}{N^2}\to\infty \quad\Longleftrightarrow\quad 4\beta-\alpha-2>0 \quad\Longleftrightarrow\quad \beta>\frac{1}{2}+\frac{\alpha}{4}. \] The computationally intensive scan statistic is \[ T_{\mathrm{scan}}=\max_{S':\,|S'|=K}\sum_{i\tau_2\), has runtime \(\Theta\!\bigl(\binom{N}{K}\bigr)\) in naive form, and succeeds when \[ K^2q \gg K\log(Ne/K), \] or, ignoring logarithms, \[ 2\beta-\alpha>0 \quad\Longleftrightarrow\quad \beta>\alpha. \] Thus the scan achieves the statistical threshold up to log factors, while the linear statistic matches the best possible polynomial-time boundary in the simple regime [1406.6625]. The computational lower bound is obtained through a randomized reduction from planted clique. Under the planted clique hypothesis—namely that planted clique detection is intractable in polynomial time for \(k=o(n^{1/2})\) at any constant \(\gamma\in(0,1/2]\)—the paper constructs a randomized mapping from \((n,k,\gamma)\) to \((N,K,p,q)\) with \(N=\ell n\), \(K=k\ell\), and \(p=2q\). The reduction preserves the null exactly and approximates the alternative in total variation, so any polynomial-time solver for the planted dense subgraph instance would transfer to a polynomial-time solver for planted clique. If the hypothesis holds for all \(\gamma\in(0,1/2]\), the efficient boundary simplifies to \[ \beta^{\sharp}(\alpha)=\frac{1}{2}+\frac{\alpha}{4}, \] and no polynomial-time algorithm achieves reliable detection when \[ \alpha<\beta<\beta^{\sharp}(\alpha). \] This is the formal source of the computational-statistical gap for \(\alpha<2/3\) [1406.6625]. Later low-degree work on planted-vs-planted testing in planted dense subgraph models identifies a sharp strong-testing threshold at \[ \lambda=(\rho\sqrt{en})^{-1}, \] achieved by counting balanced unicyclic graphs, and shows that trees are uninformative while cyclic structures carry the signal. That threshold coincides, down to the sharp constant, with the known low-degree recovery threshold [2606.05266]. This suggests that the combinatorial role of cycle-like substructures remains central even when the hypotheses are both planted rather than planted-versus-null. ## 4. Recovery, densest subgraph, and exact localization Detection asks only whether a planted anomaly is present; recovery asks for the community itself. In the sparse Bernoulli model \(p=cq\), \(q=N^{-\alpha}\), \(K=N^\beta\), exact or near-exact recovery is known to be possible information-theoretically if and only if \(\beta>\alpha\) and \(\alpha<1\). Efficient recovery is known in the region \[ \beta>\frac{1}{2}+\frac{\alpha}{2}, \] via convex relaxations or spectral or iterative methods. The same framework also yields hardness for average-case approximation of densest \(K\)-subgraph: under planted clique hardness, any constant-factor approximation is hard on average in the hard regime \[ \alpha<\beta<\frac{1}{2}+\frac{\alpha}{4}, \] whereas in the simple region \(\beta>1/2+\alpha/2\), recovering the planted community and thus obtaining a \((1+\varepsilon)\)-approximation is possible in polynomial time. The reduction extends to deterministic-size planted subgraphs for monotone tests and gives average-case computational hardness of recovery in the same hard regime [1406.6625]. Weighted models sharpen the distinction between community recovery and anomalous-subgraph recovery. In the Gaussian weighted planted dense subgraph model with planted set of size \(n/2\), exact recovery is impossible when the same \(\mathrm{SNR}<3/2\), even statistically, whereas the maximum-likelihood estimator succeeds when \(\mathrm{SNR}>2\), and the semidefinite relaxation succeeds down to the threshold value of \(2\). By contrast, in the Gaussian weighted stochastic block model with two symmetric communities, exact recovery is possible, both statistically and algorithmically, down to \(\mathrm{SNR}=1\). The paper therefore shows that exact recovery of two symmetric communities is a strictly easier problem than recovering a planted dense subgraph of size half the total number of nodes [2402.12515]. For linear-size planted dense subgraphs \(K=\Theta(n)\), however, simple spectral procedures can be optimal. In the sparse Bernoulli planted dense subgraph model with \[ p=\frac{a\log n}{n},\qquad q=\frac{b\log n}{n},\qquad K=\lfloor \rho n\rfloor, \] exact recovery is achieved by a linear-combination-of-eigenvectors spectral algorithm whenever \[ \rho f(a,b)>1, \] matching the information-theoretic threshold. In submatrix localization with \[ \mu=a\sqrt{\frac{\log n}{n}},\qquad K=\lfloor \rho n\rfloor, \] thresholding the top eigenvector achieves exact recovery whenever \[ \rho a^2>8. \] The same paper proves optimal exact recovery for a censored planted dense subgraph model via a signed-adjacency spectral algorithm under \[ t\rho\Delta_+(p,q)>1 \] [2203.11847]. ## 5. Heterogeneous, semi-random, and unbalanced graph models The homogeneous Erdős–Rényi background is mathematically clean but restrictive. In an inhomogeneous random graph \(\mathcal{G}(n,p,W)\), edges are independent conditional on a nonnegative weight vector \(W\), with \[ \mathbb{P}(A_{ij}=1\mid W)=p\,W_iW_j. \] The planted model \(\mathcal{G}\!\left(n,r,W,\frac{a}{n},\frac{b}{n}\right)\) introduces a latent Bernoulli label vector \(Z\) and within-community edge probability \(W_iW_j\,a/n\) against baseline \(W_iW_j\,b/n\). For testing \[ H_0:a=b,\qquad H_1:a>b, \] the proposed polynomial-time statistic is based on triangle and 6-cycle densities, \[ \widehat C_3,\qquad \widehat C_6,\qquad T_n=\frac{\sqrt{\binom{n}{3}\,(\widehat C_3^2-\widehat C_6)}}{2\widehat C_3\sqrt{\widehat C_3}}. \] Under mild moment bounds on \(W\), \(T_n \overset{d}{\to}\mathcal{N}(0,1)\) under \(H_0\). Under \(H_1\), \[ T_n=\lambda_n+O_p(1),\qquad \lambda_n^2=\frac{(a-b)^6b^6r^6(1-r)^6\|W\|_2^6}{n^{12}p_0^9}, \] and if \(a\asymp b\), power tends to one exactly when \[ \frac{r^2(a-b)\|W\|_2^2}{n}\to\infty. \] This supplies a parameter-free polynomial-time test for planted communities in heterogeneous networks [2101.05928]. A more general inhomogeneous framework allows arbitrary baseline probabilities \(p_{ij}\). Under the null, \(A_{ij}\sim\mathrm{Bernoulli}(p_{ij})\); under the alternative, a hidden set \(S\) of size \(k\) has \(q_{ij}=\gamma p_{ij}\) for \(i,j\in S\). The scan statistic \[ T^{(k)}(D)=\frac{E_0[e(D)]\,h\!\left(\left[\frac{e(D)}{E_0[e(D)]}-1\right]_+\right)}{|D|\log(n/|D|)},\qquad h(x)=(x+1)\log(x+1)-x, \] is maximized over \(D\subseteq[n]\), \(|D|\le k\). The information-theoretic lower bound and the upper bound of the scan test are both driven by \[ \max_{D\subseteq S}\frac{E_0[e(D)]\,h(\gamma-1)}{|D|\log(n/|D|)}, \] showing that the “most informative subgraph” inside the planted community, rather than the full community, can determine detectability in inhomogeneous graphs [1909.03217]. Other graph models change the geometry of the anomaly rather than the noise law. In the semi-random planted sparse vertex cut model, the anomaly is a pair of well-connected groups \(S\) and \(S'\) linked mainly by few connector or ambassador vertices \(T\subset S\) and \(T'\subset S'\); the relevant complexity measure is balanced vertex expansion, and semidefinite programming yields exact recovery or constant-factor bi-criteria approximation under spectral-gap and randomness conditions [1805.09747]. In the planted partition model with arbitrarily many and highly unbalanced communities, Diamond Percolation retains an edge \(ij\) when the number of common neighbors \[ W_{ij}=\#\{u\in[n]\setminus\{i,j\}:u\leftrightarrow i,\ u\leftrightarrow j\} \] satisfies \(W_{ij}\ge 2\), then returns connected components of the retained graph. Under the size-sparsity assumption \[ n^2\,\mathbb{E}[S_n^2]\,p_{\mathrm{in}}^2\,p_{\mathrm{out}}^3=o(1),\qquad p_{\mathrm{out}}=o(n^{-4/5}), \] the method achieves exact, almost exact, or weak recovery, including power-law community-size regimes [2504.01663]. ## 6. Higher-order, directed, and circular generalizations In \(d\)-uniform hypergraphs with Gaussian weights, the planted \(k\)-densest sub-hypergraph model selects a planted community \(S\subset[p]\) of size \(k\) and observes \[ Y_{i_1,\dots,i_d}= \bigl(\beta x_{i_1}\cdots x_{i_d}+Z_{i_1,\dots,i_d}\bigr)\, \mathbbm{1}_{\{i_1<\cdots