Graph sampling is the problem of choosing a node subset via sampling matrix $\mathbf{H} \in \{0,1\}^{K \times N}$ to collect samples $\mathbf{y} = \mathbf{H} \mathbf{x} \in \mathbb{R}^K$, $K < N$, so that the target signal $\mathbf{x} \in \mathbb{R}^N$ can be reconstructed in high fidelity. While sampling on undirected graphs is well studied, we propose the first sampling scheme tailored specifically for directed graphs, leveraging a previous undirected graph sampling method based on Gershgorin disc alignment (GDAS). Concretely, given a directed positive graph $\mathcal{G}^d$ specified by random-walk graph Laplacian matrix $\mathbf{L}_{rw}$, we first define reconstruction of a smooth signal $\mathbf{x}^*$ from samples $\mathbf{y}$ using graph shift variation (GSV) $\|\mathbf{L}_{rw} \mathbf{x}\|^2_2$ as a signal prior. To minimize worst-case reconstruction error of the linear system solution $\mathbf{x}^* = \mathbf{C}^{-1} \mathbf{H}^\top \mathbf{y}$ with symmetric coefficient matrix $\mathbf{C} = \mathbf{H}^\top \mathbf{H} + \mu \mathbf{L}_{rw}^\top \mathbf{L}_{rw}$, the sampling objective is to choose $\mathbf{H}$ to maximize the smallest eigenvalue $\lambda_{\min}(\mathbf{C})$ of $\mathbf{C}$. To circumvent eigen-decomposition entirely, we maximize instead a lower bound $\lambda^-_{\min}(\mathbf{S}\mathbf{C}\mathbf{S}^{-1})$ of $\lambda_{\min}(\mathbf{C})$ -- smallest Gershgorin disc left-end of a similarity transform of $\mathbf{C}$ -- via a variant of GDAS based on Gershgorin circle theorem (GCT). Experimental results show that our sampling method yields smaller signal reconstruction errors at a faster speed compared to competing schemes.
A basic premise in graph signal processing (GSP) is that a graph encoding pairwise (anti-)correlations of the targeted signal as edge weights is exploited for graph filtering. However, existing fast graph sampling schemes are designed and tested only for positive graphs describing positive correlations. In this paper, we show that for datasets with strong inherent anti-correlations, a suitable graph contains both positive and negative edge weights. In response, we propose a linear-time signed graph sampling method centered on the concept of balanced signed graphs. Specifically, given an empirical covariance data matrix $\bar{\bf{C}}$, we first learn a sparse inverse matrix (graph Laplacian) $\mathcal{L}$ corresponding to a signed graph $\mathcal{G}$. We define the eigenvectors of Laplacian $\mathcal{L}_B$ for a balanced signed graph $\mathcal{G}_B$ -- approximating $\mathcal{G}$ via edge weight augmentation -- as graph frequency components. Next, we choose samples to minimize the low-pass filter reconstruction error in two steps. We first align all Gershgorin disc left-ends of Laplacian $\mathcal{L}_B$ at smallest eigenvalue $\lambda_{\min}(\mathcal{L}_B)$ via similarity transform $\mathcal{L}_p = \S \mathcal{L}_B \S^{-1}$, leveraging a recent linear algebra theorem called Gershgorin disc perfect alignment (GDPA). We then perform sampling on $\mathcal{L}_p$ using a previous fast Gershgorin disc alignment sampling (GDAS) scheme. Experimental results show that our signed graph sampling method outperformed existing fast sampling schemes noticeably on various datasets.
Prediction of annual crop yields at a county granularity is important for national food production and price stability. In this paper, towards the goal of better crop yield prediction, leveraging recent graph signal processing (GSP) tools to exploit spatial correlation among neighboring counties, we denoise relevant features via graph spectral filtering that are inputs to a deep learning prediction model. Specifically, we first construct a combinatorial graph with edge weights that encode county-to-county similarities in soil and location features via metric learning. We then denoise features via a maximum a posteriori (MAP) formulation with a graph Laplacian regularizer (GLR). We focus on the challenge to estimate the crucial weight parameter $\mu$, trading off the fidelity term and GLR, that is a function of noise variance in an unsupervised manner. We first estimate noise variance directly from noise-corrupted graph signals using a graph clique detection (GCD) procedure that discovers locally constant regions. We then compute an optimal $\mu$ minimizing an approximate mean square error function via bias-variance analysis. Experimental results from collected USDA data show that using denoised features as input, performance of a crop yield prediction model can be improved noticeably.
In the graph signal processing (GSP) literature, graph Laplacian regularizer (GLR) was used for signal restoration to promote smooth reconstructions with respect to the underlying graph -- typically signals that are (piecewise) constant. However, for graph signals that are (piecewise) planar, GLR may suffer from the well-known "staircase" effect. In this paper, focusing on manifold graphs -- sets of uniform discrete samples on low-dimensional continuous manifolds -- we generalize GLR to gradient graph Laplacian regularizer (GGLR) that provably promotes piecewise planar (PWP) signal reconstruction. Specifically, for a graph endowed with latent space coordinates (e.g., 2D images, 3D point clouds), we first define a gradient operator, using which we construct a higher-order gradient graph for the computed gradients in each latent dimension. This maps to a gradient-induced nodal graph (GNG) and a Laplacian matrix for a signed graph that is provably positive semi-definite (PSD), thus suitable for quadratic regularization. For manifold graphs without explicit latent coordinates, we propose a fast parameter-free spectral method to first compute latent space coordinates for graph nodes based on generalized eigenvectors. We derive the means-square-error minimizing weight parameter for GGLR efficiently, trading off bias and variance of the signal estimate. Experimental results show that GGLR outperformed previous graph signal priors like GLR and graph total variation (GTV) in a range of graph signal restoration tasks.
Transform coding to sparsify signal representations remains crucial in an image compression pipeline. While the Karhunen-Lo\`{e}ve transform (KLT) computed from an empirical covariance matrix $\bar{C}$ is theoretically optimal for a stationary process, in practice, collecting sufficient statistics from a non-stationary image to reliably estimate $\bar{C}$ can be difficult. In this paper, to encode an intra-prediction residual block, we pursue a hybrid model-based / data-driven approach: the first $K$ eigenvectors of a transform matrix are derived from a statistical model, e.g., the asymmetric discrete sine transform (ADST), for stability, while the remaining $N-K$ are computed from $\bar{C}$ for performance. The transform computation is posed as a graph learning problem, where we seek a graph Laplacian matrix minimizing a graphical lasso objective inside a convex cone sharing the first $K$ eigenvectors in a Hilbert space of real symmetric matrices. We efficiently solve the problem via augmented Lagrangian relaxation and proximal gradient (PG). Using WebP as a baseline image codec, experimental results show that our hybrid graph transform achieved better energy compaction than default discrete cosine transform (DCT) and better stability than KLT.
It is now known that the expressive power of graph convolutional neural nets (GCN) does not grow infinitely with the number of layers. Instead, the GCN output approaches a subspace spanned by the first eigenvector of the normalized graph Laplacian matrix with the convergence rate characterized by the "eigen-gap": the difference between the Laplacian's first two distinct eigenvalues. To promote a deeper GCN architecture with sufficient expressiveness, in this paper, given an empirical covariance matrix $\bar{C}$ computed from observable data, we learn a sparse graph Laplacian matrix $L$ closest to $\bar{C}^{-1}$ while maintaining a desirable eigen-gap that slows down convergence. Specifically, we first define a sparse graph learning problem with constraints on the first eigenvector (the most common signal) and the eigen-gap. We solve the corresponding dual problem greedily, where a locally optimal eigen-pair is computed one at a time via a fast approximation of a semi-definite programming (SDP) formulation. The computed $L$ with the desired eigen-gap is normalized spectrally and used for supervised training of GCN for a targeted task. Experiments show that our proposal produced deeper GCNs and smaller errors compared to a competing scheme without explicit eigen-gap optimization.
Our goal is to efficiently compute low-dimensional latent coordinates for nodes in an input graph -- known as graph embedding -- for subsequent data processing such as clustering. Focusing on finite graphs that are interpreted as uniformly samples on continuous manifolds (called manifold graphs), we leverage existing fast extreme eigenvector computation algorithms for speedy execution. We first pose a generalized eigenvalue problem for sparse matrix pair $(\A,\B)$, where $\A = \L - \mu \Q + \epsilon \I$ is a sum of graph Laplacian $\L$ and disconnected two-hop difference matrix $\Q$. Eigenvector $\v$ minimizing Rayleigh quotient $\frac{\v^{\top} \A \v}{\v^{\top} \v}$ thus minimizes $1$-hop neighbor distances while maximizing distances between disconnected $2$-hop neighbors, preserving graph structure. Matrix $\B = \text{diag}(\{\b_i\})$ that defines eigenvector orthogonality is then chosen so that boundary / interior nodes in the sampling domain have the same generalized degrees. $K$-dimensional latent vectors for the $N$ graph nodes are the first $K$ generalized eigenvectors for $(\A,\B)$, computed in $\cO(N)$ using LOBPCG, where $K \ll N$. Experiments show that our embedding is among the fastest in the literature, while producing the best clustering performance for manifold graphs.
A 3D point cloud is typically constructed from depth measurements acquired by sensors at one or more viewpoints. The measurements suffer from both quantization and noise corruption. To improve quality, previous works denoise a point cloud \textit{a posteriori} after projecting the imperfect depth data onto 3D space. Instead, we enhance depth measurements directly on the sensed images \textit{a priori}, before synthesizing a 3D point cloud. By enhancing near the physical sensing process, we tailor our optimization to our depth formation model before subsequent processing steps that obscure measurement errors. Specifically, we model depth formation as a combined process of signal-dependent noise addition and non-uniform log-based quantization. The designed model is validated (with parameters fitted) using collected empirical data from an actual depth sensor. To enhance each pixel row in a depth image, we first encode intra-view similarities between available row pixels as edge weights via feature graph learning. We next establish inter-view similarities with another rectified depth image via viewpoint mapping and sparse linear interpolation. This leads to a maximum a posteriori (MAP) graph filtering objective that is convex and differentiable. We optimize the objective efficiently using accelerated gradient descent (AGD), where the optimal step size is approximated via Gershgorin circle theorem (GCT). Experiments show that our method significantly outperformed recent point cloud denoising schemes and state-of-the-art image denoising schemes, in two established point cloud quality metrics.
We study the problem of efficiently summarizing a short video into several keyframes, leveraging recent progress in fast graph sampling. Specifically, we first construct a similarity path graph (SPG) $\mathcal{G}$, represented by graph Laplacian matrix $\mathbf{L}$, where the similarities between adjacent frames are encoded as positive edge weights. We show that maximizing the smallest eigenvalue $\lambda_{\min}(\mathbf{B})$ of a coefficient matrix $\mathbf{B} = \text{diag}(\mathbf{a}) + \mu \mathbf{L}$, where $\mathbf{a}$ is the binary keyframe selection vector, is equivalent to minimizing a worst-case signal reconstruction error. We prove that, after partitioning $\mathcal{G}$ into $Q$ sub-graphs $\{\mathcal{G}^q\}^Q_{q=1}$, the smallest Gershgorin circle theorem (GCT) lower bound of $Q$ corresponding coefficient matrices -- $\min_q \lambda^-_{\min}(\mathbf{B}^q)$ -- is a lower bound for $\lambda_{\min}(\mathbf{B})$. This inspires a fast graph sampling algorithm to iteratively partition $\mathcal{G}$ into $Q$ sub-graphs using $Q$ samples (keyframes), while maximizing $\lambda^-_{\min}(\mathbf{B}^q)$ for each sub-graph $\mathcal{G}^q$. Experimental results show that our algorithm achieves comparable video summarization performance as state-of-the-art methods, at a substantially reduced complexity.
Sensor placement for linear inverse problems is the selection of locations to assign sensors so that the entire physical signal can be well recovered from partial observations. In this paper, we propose a fast sampling algorithm to place sensors. Specifically, assuming that the field signal $\mathbf{f}$ is represented by a linear model $\mathbf{f}=\pmb{\phi}\mathbf{g}$, it can be estimated from partial noisy samples via an unbiased least-squares (LS) method, whose expected mean square error (MSE) depends on chosen samples. First, we formulate an approximate MSE problem, and then prove it is equivalent to a problem related to a principle submatrix of $\pmb{\phi}\pmb{\phi}^\top$ indexed by sample set. To solve the formulated problem, we devise a fast greedy algorithm with simple matrix-vector multiplications, leveraging a matrix inverse formula. To further reduce complexity, we reuse results in the previous greedy step for warm start, so that candidates can be evaluated via lightweight vector-vector multiplications. Extensive experiments show that our proposed sensor placement method achieved the lowest sensor sampling time and the best performance compared to state-of-the-art schemes.