Fact-checked by Grok 2 weeks ago

Decision boundary

In , particularly in supervised tasks, a decision boundary is a that partitions the input feature space into regions, each associated with a specific label, enabling the classifier to assign predictions based on which region a data point falls into. This boundary represents the set of points where the model's output transitions between classes, often corresponding to equal posterior probabilities for the involved classes. The nature of the decision boundary varies depending on the complexity of the data and the algorithm used; linear boundaries, which are straight lines in two dimensions or hyperplanes in higher dimensions, are characteristic of simple models like and linear support vector machines (SVMs), where the boundary is defined by an such as w^T x + b = 0, with [w](/page/W) as the normal and b as the term. In contrast, nonlinear boundaries—curved surfaces or more intricate hypersurfaces—are essential for capturing complex, non-separable data distributions, such as overlapping clusters, and are achieved through techniques like kernel methods in SVMs, decision trees, or deep neural networks. The position and smoothness of the decision are critical for a classifier's performance, as a boundary that closely hugs the training data without enhances robustness to unseen examples, while overly complex boundaries can lead to poor . For instance, in SVMs, the optimal boundary maximizes the margin—the distance to the nearest training points (support vectors)—to improve classification reliability. Visualizing decision boundaries in low-dimensional spaces aids in evaluating model effectiveness, but in high dimensions, their analysis often relies on or boundary approximation methods to assess model behavior.

Fundamentals

Definition

In , particularly within supervised tasks, a decision boundary refers to the that partitions the feature space into regions corresponding to different class labels, such that points on one side of the boundary are predicted to belong to one class while points on the other side are assigned to another. This boundary delineates the separator where the model's predictions transition from one class to another, enabling the classifier to make discrete decisions based on the input features. Decision boundaries can be characterized as hard or soft depending on the classifier's formulation. A hard boundary enforces a strict separation, assuming the data is perfectly separable without errors, as in the case of linearly separable datasets where no training points violate the . In contrast, soft boundaries accommodate noise or overlapping classes by allowing some margin violations or incorporating probabilistic outputs, where the boundary represents the locus of points with equal posterior probabilities for the classes, such as a probability of 0.5 in binary . This probabilistic interpretation is common in models that output confidence scores rather than binary decisions. The concept plays a central role in supervised learning for both binary and multiclass classification problems. In binary classification, the boundary divides the space into two regions, one for each class; for multiclass scenarios with K > 2 classes, it generalizes to multiple hypersurfaces or a more complex partitioning that assigns regions to each of the K classes, often constructed via pairwise or one-versus-all strategies. For instance, in a simple two-dimensional binary classification problem involving features like height and weight to distinguish between two groups, the decision boundary might manifest as a straight line separating the clusters of data points from each class.

Mathematical Representation

In , the decision boundary is mathematically formalized as the set of points in the feature space where the classifier's output equals a , typically zero, marking the transition between class predictions. For a general classifier with decision f(\mathbf{x}), the boundary is defined by \{ \mathbf{x} \in \mathbb{R}^d \mid f(\mathbf{x}) = 0 \}, separating regions assigned to different classes. In , the decision often takes the form f(\mathbf{x}) = 0, where the sign of f(\mathbf{x}) determines the class label. A foundational linear case is given by the f(\mathbf{x}) = \mathbf{w}^T \mathbf{x} + b = 0, with \mathbf{w} as the weight to the and b as the term shifting it from the ; points satisfying this lie on the , while f(\mathbf{x}) > 0 or f(\mathbf{x}) < 0 assigns them to one class or the other. For multiclass problems with K > 2 classes, the decision boundary extends via pairwise separations or probabilistic assignments. In the one-vs-all approach, K binary boundaries are defined, each as f_k(\mathbf{x}) = 0 for class k versus the rest, with the overall assignment to \arg\max_k f_k(\mathbf{x}). Alternatively, using softmax outputs p(C_k \mid \mathbf{x}) = \frac{\exp(a_k)}{\sum_{j=1}^K \exp(a_j)}, where a_k = \mathbf{w}_k^T \mathbf{x} + b_k, the boundary between classes k and j occurs where p(C_k \mid \mathbf{x}) = p(C_j \mid \mathbf{x}) > p(C_m \mid \mathbf{x}) for all other m, yielding hypersurfaces from a_k = a_j. The decision regions are the connected components of the feature space partitioned by these boundaries, each corresponding to the set R_k = \{ \mathbf{x} \mid \hat{y}(\mathbf{x}) = k \}, where \hat{y}(\mathbf{x}) is the predicted ; these regions may be for linear boundaries but arbitrary otherwise.

Linear Decision Boundaries

In Linear Classifiers

In linear classifiers, the decision boundary is defined as a hyperplane that separates data points of different classes in the feature space. This hyperplane is mathematically represented by the equation \mathbf{w}^T \mathbf{x} + b = 0, where \mathbf{w} is the weight vector normal to the hyperplane, \mathbf{x} is the input feature vector, and b is the bias term that shifts the hyperplane from the origin. Points on one side of the hyperplane (\mathbf{w}^T \mathbf{x} + b > 0) are classified into one class, while those on the other side (\mathbf{w}^T \mathbf{x} + b < 0) belong to the opposite class. This linear separation assumes that the classes are distinguishable by a straight line in two dimensions or a plane in higher dimensions. The perceptron, introduced by Rosenblatt in 1958, exemplifies this linear decision boundary in binary classification tasks. It computes the weighted sum \mathbf{w} \cdot \mathbf{x} and applies a step function to produce a hard classification, with the boundary forming where the sum equals the threshold (effectively b = 0 in normalized form). The perceptron algorithm iteratively adjusts \mathbf{w} to converge on this hyperplane only if the data is linearly separable, meaning no points from one class lie on the wrong side of the boundary. In contrast, logistic regression extends this framework by incorporating a probabilistic interpretation through the sigmoid activation function, \sigma(z) = \frac{1}{1 + e^{-z}}, where z = \mathbf{w}^T \mathbf{x} + b. The predicted probability of class 1 is \sigma(z), and the decision boundary remains the linear hyperplane where \sigma(z) = 0.5, or equivalently z = 0. This soft boundary allows for confidence scores rather than binary outputs, making it suitable for applications requiring probability estimates. Linear classifiers rely on the assumption of linear separability, where two classes can be perfectly divided by a hyperplane without overlap. If the data satisfies this condition—such as two convex, non-intersecting sets—the perceptron guarantees convergence to an optimal boundary. However, real-world data often violates this assumption due to noise or overlapping distributions, leading to misclassifications or divergence in unregularized models. To address non-separable data, regularization is introduced, typically L2 (ridge) penalty, which adds a term \lambda \|\mathbf{w}\|_2^2 to the loss function. This constrains the magnitude of \mathbf{w}, preventing overfitting and stabilizing the decision boundary by favoring simpler hyperplanes that generalize better, even if they allow some training errors. In logistic regression, the regularized loss becomes L(\mathbf{w}) = -\frac{1}{n} \sum_{i=1}^n [y_i \log(\hat{p}_i) + (1 - y_i) \log(1 - \hat{p}_i)] + \lambda \|\mathbf{w}\|_2^2, where \hat{p}_i = \sigma(\mathbf{w}^T \mathbf{x}_i + b). The decision boundary in logistic regression is refined through gradient descent optimization of the negative log-likelihood (cross-entropy) loss. The loss function is E(\mathbf{w}) = -\sum_{n=1}^N [t_n \ln y_n + (1 - t_n) \ln(1 - y_n)], where t_n is the target label (0 or 1), and y_n = \sigma(\mathbf{w}^T \mathbf{x}_n + b). The gradient with respect to \mathbf{w} is \nabla E(\mathbf{w}) = \sum_{n=1}^N (y_n - t_n) \mathbf{x}_n, derived by applying the chain rule to the sigmoid derivative \sigma'(z) = \sigma(z)(1 - \sigma(z)). The update rule for each iteration is \mathbf{w}^{(k+1)} = \mathbf{w}^{(k)} - \alpha \nabla E(\mathbf{w}^{(k)}), where \alpha is the learning rate; a similar update applies to b. This process shifts the hyperplane toward minimizing prediction errors, effectively adjusting the boundary to better separate classes by increasing the margin between misclassified points and the current hyperplane. For the bias, the update is b^{(k+1)} = b^{(k)} - \alpha \sum_{n=1}^N (y_n - t_n). With regularization, the gradient includes an additional term $2\lambda \mathbf{w}, ensuring the boundary remains robust for non-separable cases.

In Support Vector Machines

In support vector machines (SVMs), the decision boundary is defined as the optimal hyperplane that separates two classes of data points while maximizing the geometric margin between them, providing a robust linear classifier less prone to overfitting compared to other linear methods. The hyperplane is represented as \mathbf{w} \cdot \mathbf{x} + b = 0, where \mathbf{w} is the normal vector to the hyperplane and b is the bias term; the goal is to maximize the margin, which is given by $2 / \|\mathbf{w}\|, achieved by minimizing \|\mathbf{w}\|^2 / 2 subject to the constraints y_i (\mathbf{w} \cdot \mathbf{x}_i + b) \geq 1 for all training points \mathbf{x}_i with labels y_i \in \{-1, 1\}. This formulation ensures that points are classified correctly with a buffer zone on either side of the hyperplane, where the boundaries of the margin are \mathbf{w} \cdot \mathbf{x} + b = \pm 1. The optimization in SVMs is closely tied to the hinge loss function, which penalizes misclassifications and points within the margin: \max(0, 1 - y_i f(\mathbf{x}_i)), where f(\mathbf{x}_i) = \mathbf{w} \cdot \mathbf{x}_i + b. This loss encourages the decision boundary to position itself such that the functional margin y_i f(\mathbf{x}_i) is at least 1 for all points, directly influencing the placement of the hyperplane to maximize separation while controlling the trade-off between margin size and classification errors. For linearly separable data, the hard-margin SVM assumes perfect separation, but real-world datasets often overlap, necessitating soft margins to allow some violations. Soft margins introduce non-negative slack variables \xi_i \geq 0 for each point, relaxing the constraints to y_i (\mathbf{w} \cdot \mathbf{x}_i + b) \geq 1 - \xi_i, and the objective becomes minimizing \|\mathbf{w}\|^2 / 2 + C \sum \xi_i, where C > 0 is a regularization parameter controlling the penalty for margin violations. Points with \xi_i > 0 lie within or on the wrong side of the margin, enabling the algorithm to find a feasible boundary even for non-separable cases. The dual formulation of the SVM optimization problem provides an efficient way to compute the decision boundary, transforming the primal problem into maximizing \sum_i \alpha_i - \frac{1}{2} \sum_{i,j} \alpha_i \alpha_j y_i y_j (\mathbf{x}_i \cdot \mathbf{x}_j) subject to \sum_i \alpha_i y_i = 0 and $0 \leq \alpha_i \leq C, where \alpha_i are Lagrange multipliers. This dual form reveals the support vectors—those training points with \alpha_i > 0—which are the critical points closest to or crossing the margin that alone define the position of the , as \mathbf{w} = \sum_i \alpha_i y_i \mathbf{x}_i. Only these support vectors influence the decision boundary, making SVMs sparse and computationally efficient for large datasets.

Nonlinear Decision Boundaries

Kernel Methods

Kernel methods enable linear classifiers, such as support vector machines, to produce nonlinear decision boundaries by implicitly mapping input data into a higher-dimensional feature space. The kernel trick achieves this through a feature map \phi: \mathcal{X} \to \mathcal{H}, where \mathcal{H} is a , transforming the decision boundary from the original input space to \mathbf{w} \cdot \phi(\mathbf{x}) + b = 0 in the feature space. This mapping allows the model to separate nonlinearly separable data without explicitly computing the potentially infinite-dimensional \phi(\mathbf{x}), instead relying on the function K(\mathbf{x}, \mathbf{z}) = \phi(\mathbf{x}) \cdot \phi(\mathbf{z}) to evaluate inner products efficiently. Common kernel functions define the geometry of this implicit feature space and thus shape the resulting nonlinear boundary. The , K(\mathbf{x}, \mathbf{z}) = (\mathbf{x} \cdot \mathbf{z} + c)^d, generates boundaries resembling polynomial surfaces of degree d, suitable for data with moderate curvature, where c \geq 0 controls the influence of higher-order terms and d determines complexity. The (RBF) kernel, K(\mathbf{x}, \mathbf{z}) = \exp(-\gamma \|\mathbf{x} - \mathbf{z}\|^2), produces highly flexible, localized boundaries that can capture intricate patterns like clusters or spirals, with \gamma > 0 tuning the width of the Gaussian radial basis. These are widely used due to their , ensuring a valid inner product in \mathcal{H}. The formalizes how the decision function is constructed solely from kernel evaluations on points, stating that the solution to the minimizes the regularized and takes the form f(\mathbf{x}) = \sum_{i=1}^n \alpha_i y_i K(\mathbf{x}_i, \mathbf{x}) + b, where \alpha_i are Lagrange multipliers, y_i are labels, and only support vectors (where \alpha_i > 0) contribute. This expansion confines the boundary's complexity to the training data's span in feature space, promoting sparsity. However, kernel methods incur trade-offs: via scales as O(n^3) in the number of samples n due to kernel inversion, limiting scalability to large datasets, while high-dimensional mappings , especially with flexible kernels like RBF, necessitating regularization via parameters such as the margin softener C.

Neural Networks

In neural networks, the decision boundary is formed through a of affine transformations and nonlinear functions applied across multiple layers. For a network with L hidden layers, the output function can be expressed as \mathbf{f}(\mathbf{x}) = \sigma_L \left( W_L \sigma_{L-1} \left( \dots \sigma_1 (W_1 \mathbf{x} + \mathbf{b}_1) \dots \right) + \mathbf{b}_L \right), where \mathbf{x} is the input , W_i and \mathbf{b}_i are the weight matrices and vectors for layer i, and \sigma_i denotes the (e.g., or ReLU) at that layer. This layered structure transforms the input space iteratively, with each layer's affine operation W_i \mathbf{h}_{i-1} + \mathbf{b}_i followed by a nonlinearity \sigma_i that introduces bends and folds into the boundary, enabling the network to separate complex data distributions. The nature of the decision boundary evolves significantly with network depth and activation choices. A single-layer produces a linear decision boundary, akin to those in basic linear classifiers, but stacking multiple layers with nonlinear activations like the \sigma(z) = \frac{1}{1 + e^{-z}} or the rectified linear unit (ReLU) \sigma(z) = \max(0, z) generates highly nonlinear boundaries capable of capturing intricate patterns. ReLU, in particular, promotes sparsity and faster convergence during training, leading to piecewise linear boundaries that approximate smooth nonlinearities in deeper architectures. The universal approximation theorem establishes that neural networks with a single hidden layer and sufficient neurons using nonlinear activations like can approximate any on a compact subset of \mathbb{R}^n to arbitrary accuracy, implying their ability to form decision boundaries for any separable continuous data distribution. This capability extends to deeper networks, which often achieve better approximations for high-dimensional problems. During training, the decision boundary is optimized via , an algorithm that computes gradients of a (e.g., ) with respect to the weights by propagating errors backward through the layers using the chain rule, iteratively adjusting parameters to minimize misclassifications and refine the boundary.

Visualization and Analysis

Plotting Techniques

Plotting techniques for decision boundaries typically involve evaluating a classifier over a discretized of the input space to generate visualizations that reveal how the model separates . This -based evaluation method creates a using functions like NumPy's meshgrid, which generates coordinate for each , allowing the classifier to predict labels or probabilities at each grid point. The resulting predictions form a or that can be contoured to delineate boundaries where transitions occur, providing insights into model behavior for and . In practice, libraries such as offer dedicated tools like DecisionBoundaryDisplay for streamlined visualization, which internally handles grid creation with a default of 100 points per and supports methods like from_estimator to fit and plot boundaries directly from a trained model. This uses Matplotlib's contourf to render filled plots, where regions of uniform color represent areas assigned to specific es, and lines from contour can overlay exact boundaries. For extensions, similar contouring can be applied using plot_surface or isosurfaces, though via tools like may enhance exploration. Matplotlib's contourf is particularly versatile, accepting a grid of prediction values and mapping them to colors via colormaps, making it a foundational primitive for custom decision surface plots in binary or multiclass settings. Visualizing decision boundaries in high-dimensional spaces, where direct gridding becomes computationally infeasible beyond 3-4 dimensions, requires techniques to project data into a lower-dimensional before applying standard plotting methods. (PCA) linearly transforms features to capture maximum variance, enabling 2D or 3D slices of the boundary, though it may oversimplify nonlinear structures. t-Distributed Stochastic Neighbor Embedding (t-SNE) preserves local neighborhoods nonlinearly, often yielding clearer visualizations of compact decision regions in complex datasets by focusing on pairwise similarities. These projections allow grid-based evaluation on the reduced space, but care must be taken as distortions can alter perceived boundary smoothness; studies evaluating 28 reduction methods on datasets like found t-SNE and UMAP superior for minimizing visualization errors in boundary clarity. A representative example is the XOR problem, a classic nonlinearly separable with four points in space, where linear classifiers fail but neural networks succeed by forming boundaries. Visualizations of trained multilayer perceptrons on XOR reveal wiggly, non-convex boundaries that enclose diagonally opposite points, demonstrating how layers enable separations; such plots, generated via grid evaluation and contouring, highlight the transition from simple lines to intricate curves as network depth increases.

Interpretability Challenges

In models, decision boundaries often exhibit extreme , manifesting as -like structures that are highly non-intuitive and difficult to comprehend due to their intricate, self-similar patterns across multiple scales. This dimensionality arises from the high-capacity nature of neural networks, which can produce boundaries with varying local geometries that defy simple geometric descriptions, complicating efforts to understand globally. Such hinders interpretability, as human analysts struggle to discern underlying logic without resorting to approximations or visualizations that may oversimplify the true structure. Overfitting is a key interpretability challenge, where decision boundaries become overly jagged and convoluted to fit in the training data, leading to poor and misleading insights into the model's true capabilities. These irregular boundaries, characterized by sharp oscillations and disconnected components, serve as indicators of memorization rather than learning meaningful patterns, making it challenging to trust predictions in unseen scenarios. techniques have been proposed to mitigate this by regularizing the boundary to reduce jaggedness, thereby enhancing both interpretability and performance. To quantify these issues, researchers employ metrics such as boundary curvature and smoothness, often computed via the to assess second-order variations along the , revealing how sharply the model transitions between classes. Additional measures include classification oscillation, which tracks rapid label changes near the to gauge complexity, and approximations that estimate the 's geometric intricacy relative to the input space. Robustness to perturbations is evaluated through margin distances—the separation between points and the —where thinner margins indicate vulnerability to small input changes, underscoring interpretability gaps in adversarial settings. Thick boundaries, conversely, correlate with improved robustness and smoother interpretations. Local interpretability techniques like and SHAP address these challenges by approximating decision boundaries around specific instances, providing feature attributions that elucidate how perturbations near a point influence classification. fits interpretable surrogate models, such as linear classifiers, to sampled perturbations, yielding localized linear approximations of nonlinear boundaries for intuitive explanations. SHAP, grounded in , assigns importance values to features based on their marginal contributions to boundary crossings, offering consistent local insights even for complex global structures. These methods, while effective for point-specific analysis, reveal the broader difficulty in scaling explanations to the entire boundary without losing fidelity.

Applications

Synthetic Examples

Synthetic examples employ controlled, low-dimensional toy datasets to demonstrate the characteristics of decision boundaries in models, highlighting how linear and nonlinear classifiers perform on data with varying separability. These datasets are particularly useful for pedagogical purposes, as they allow clear illustration of boundary shapes without the complications of real-world noise or high dimensionality. The Iris dataset, introduced by in 1936, consists of 150 samples from three species of iris flowers, each described by four features: sepal length, sepal width, petal length, and petal width. In two-dimensional projections, such as petal length versus petal width, the dataset exhibits linear separability, particularly for distinguishing from the other two species (versicolor and virginica), where a straight line can effectively separate the classes with minimal overlap. Linear classifiers like or linear support vector machines achieve high accuracy, often exceeding 95%, on these projections due to the clear linear boundaries. The XOR problem, a classic task with inputs (0,0) → 0, (0,1) → 1, (1,0) → 1, and (1,1) → 0, exemplifies data that is not linearly separable in two dimensions, as no single straight line can partition the points correctly. and demonstrated in their 1969 book Perceptrons that single-layer linear classifiers, such as the , fail on this , achieving at most 50% accuracy by misclassifying at least two points. However, nonlinear models succeed: a two-layer or a with a quadratic kernel can learn the required piecewise linear boundary, attaining 100% accuracy on the four training points. The half-moon dataset, a synthetic problem generated as two interleaving semicircles, requires curved decision to separate the classes, making it linearly inseparable in the original . Models using (RBF) kernels in support vector machines map the data to a higher-dimensional where a linear boundary becomes feasible, resulting in near-perfect separation and accuracy close to 100%. Similarly, deep neural networks with nonlinear activations can approximate the curved boundary effectively, also achieving high performance.
DatasetLinear Classifier PerformanceNonlinear Model (e.g., RBF or 2-Layer ) PerformanceTypical Boundary Shape
(2D projection)~95% accuracy (e.g., )~98% accuracy (e.g., SVM with )Straight line for separable classes
XOR50% accuracy (e.g., )100% accuracy (e.g., kernel SVM)Piecewise linear
Half-moon~50% accuracy (e.g., linear SVM)~100% accuracy (e.g., RBF SVM)Curved (sigmoidal or circular)
These performances are based on standard implementations and evaluations on the full datasets, with linear models struggling on nonlinearly separable cases while kernel methods and neural networks adapt via implicit feature transformations.

Real-World Use Cases

In image classification tasks, convolutional neural networks (CNNs) form complex decision boundaries in high-dimensional or spaces to distinguish between classes, such as handwritten digits in the MNIST dataset. For instance, a on MNIST reveals that decision boundaries evolve during optimization, with the minimum distance from data points to the boundary increasing as the model learns more separable representations, achieving error rates below 1% on test sets. These boundaries are often nonlinear and intricate, capturing subtle variations like stroke thickness or orientation that separate digits like 4 and 9. further evaluates such boundaries on MNIST, showing that metrics can quantify boundary complexity and aid model selection for improved . In , decision boundaries are constructed in spaces to predict diseases from physiological indicators, enabling precise with limited samples. Support vector machines (SVMs), for example, define hyperplanes that separate healthy and diseased states based on features like tumor markers or blood metabolites in , where kernel tricks allow nonlinear boundaries to achieve accuracies exceeding 95% on clinical datasets. Neural networks extend this by learning hierarchical boundaries from multi-omics data, such as profiles for prediction, where boundaries in space highlight critical thresholds for intervention. These approaches prioritize margin maximization to handle noisy clinical data, improving diagnostic reliability over traditional thresholds. Anomaly detection employs one-class decision boundaries to encircle normal data distributions, identifying outliers as points falling outside this region without requiring labeled anomalies. The one-class SVM, a foundational method, fits a or around the majority class in high-dimensional spaces, such as network traffic or sensor readings, where the boundary's radius determines anomaly scores based on distance metrics. In practice, this has been applied to detection in financial transactions, achieving detection rates of 90% or higher by enclosing legitimate patterns while flagging deviations. Deep variants, like one-class neural networks, refine these boundaries for image or time-series data, enhancing robustness in settings. Real-world decision boundaries face significant challenges from high dimensionality and class imbalance, which can distort reliability and . In high-dimensional spaces, such as genomic biomarkers, the curse of dimensionality leads to sparse data distributions, making boundary estimation prone to and requiring regularization techniques like Barron norms for approximation rates that scale polynomially with dimensions. Class imbalance exacerbates this by shifting boundaries toward minority classes during , causing distortion near the edge where down-sampling majority instances oversimplifies separations and reduces minority recall. These issues are particularly acute in medical applications, where imbalanced datasets (e.g., rare diseases) demand cost-sensitive learning to maintain integrity without extensive resampling.

References

  1. [1]
    Decision Boundary - an overview | ScienceDirect Topics
    A decision boundary is defined as a hypersurface used by pattern classifiers to classify input patterns into different regions based on certain criteria.
  2. [2]
  3. [3]
    [PDF] Machine Learning What is a good Decision Boundary?
    Let w denote a vector orthogonal to the decision boundary, and b denote a scalar. "offset" term, then we can write the decision boundary as: Class 1. Class 2.Missing: definition | Show results with:definition
  4. [4]
    Decision Boundaries – Machine Learning
    A decision boundary separates two or more classes from one another. The simplest decision boundaries are straight, but it is possible to draw very complicated ...<|control11|><|separator|>
  5. [5]
    [2206.01515] Understanding Deep Learning via Decision Boundary
    Jun 3, 2022 · This paper discovers that the neural network with lower decision boundary (DB) variability has better generalizability.
  6. [6]
    [PDF] 8 Classification - cs.Toronto
    The purpose of the decision boundary is to identity the regions of the input space that correspond to each class. For binary classification the decision ...<|control11|><|separator|>
  7. [7]
    [PDF] Support-Vector Networks - Semantic Scholar
    Vapnik. Computer Science. COLT '92. 1992. A training algorithm that maximizes the margin between the training patterns and the decision boundary is presented.
  8. [8]
  9. [9]
    [PDF] Pattern Recognition and Machine Learning - Microsoft
    This new textbook reflects these recent developments while providing a compre- hensive introduction to the fields of pattern recognition and machine learning.
  10. [10]
    [PDF] Linear Classification - Statistics & Data Science
    The decision boundary of the above classifier can be characterized by the set {x ∈ X : δ1(x) = δ0(x)}, which is quadratic so this procedure is called quadratic ...
  11. [11]
    [PDF] Machine Learning I Lecture 06 Linear Separability
    If a dataset is linearly separable, then you are guaranteed to find a perfect classifier. Then you can say how good is the classifier you designed compared to ...
  12. [12]
    The Perceptron: A Probabilistic Model for Information Storage and ...
    No information is available for this page. · Learn why
  13. [13]
    [PDF] CSC 411: Lecture 04: Logistic Regression
    An alternative: replace the sign(·) with the sigmoid or logistic function. We assumed a particular functional form: sigmoid applied to a linear.
  14. [14]
    23 Logistic Regression II (Summer 2025) - Data 100
    1 Regularized Logistic Regression. To avoid large weights and infinite loss (particularly on linearly separable data), we use regularization. The same ...
  15. [15]
    [PDF] Chapter 6 - Logistic Regression
    The key difference is that we derive and optimize this decision boundary based on a likelihood model. 6.2 Training a Logistic Regression Model. Logistic ...
  16. [16]
    Support-vector networks | Machine Learning
    About this article. Cite this article. Cortes, C., Vapnik, V. Support-vector networks. Mach Learn 20, 273–297 (1995). https://doi.org/10.1007/BF00994018.
  17. [17]
    [PDF] A Training Algorithm for Optimal Margin Classi ers
    A training algorithm that maximizes the mar- gin between the training patterns and the de- cision boundary is presented. The technique.
  18. [18]
    [PDF] Learning with Kernels
    ery learning algorithm has been redesigned to exploit the power of kernel meth- ods. Bernhard Schölkopf and Alexander Smola have written a comprehensive, yet.
  19. [19]
    Support Vector Machines and Kernels for Computational Biology - NIH
    The polynomial kernel of degree 1 leads to a linear separation (A). Higher-degree polynomial kernels allow a more flexible decision boundary (B,C). The style ...Missing: seminal | Show results with:seminal
  20. [20]
    [PDF] A Generalized Representer Theorem - Alex Smola
    Abstract. Wahba's classical representer theorem states that the solu- tions of certain risk minimization problems involving an empirical risk.
  21. [21]
    Time Complexity Analysis of Support Vector Machines (SVM) in ...
    The results of our research has proved that the complexity of SVM (LibSVM) is O(n3) and the time complexity shown that C++ faster than Java.<|control11|><|separator|>
  22. [22]
    [PDF] Support Vector Machines and Kernel Methods
    risk of “overfitting.” The choice of the kernel function is the only major design choice. How- ever, this choice can be guided by domain expertise, keeping ...
  23. [23]
    Learning representations by back-propagating errors - Nature
    Oct 9, 1986 · We describe a new learning procedure, back-propagation, for networks of neurone-like units. The procedure repeatedly adjusts the weights of the connections in ...
  24. [24]
    [PDF] Rectified Linear Units Improve Restricted Boltzmann Machines
    Nair, V. and Hinton, G. E. Implicit mixtures of restricted boltzmann machines. In Neural information processing systems, 2008. Salakhutdinov, R. and ...Missing: original | Show results with:original
  25. [25]
    Approximation by superpositions of a sigmoidal function
    Feb 17, 1989 · The paper discusses approximation properties of other possible types of nonlinearities that might be implemented by artificial neural networks.
  26. [26]
    DecisionBoundaryDisplay — scikit-learn 1.7.2 documentation
    Decisions boundary visualization. It is recommended to use from_estimator to create a DecisionBoundaryDisplay. All parameters are stored as attributes.
  27. [27]
  28. [28]
    Plot classification boundaries with different SVM Kernels - Scikit-learn
    In this example, we compare the most common kernel types of Support Vector Machines: the linear kernel ( "linear" ), the polynomial kernel ( "poly" ), the ...
  29. [29]
    matplotlib.pyplot.contourf — Matplotlib 3.10.7 documentation
    `matplotlib.pyplot.contourf` plots filled contours, using height values over which the contour is drawn, and fills intervals closed at the top.Missing: decision | Show results with:decision
  30. [30]
    Constructing and Visualizing High-Quality Classifier Decision ...
    Visualizing decision boundaries of machine learning classifiers can help in classifier design, testing and fine-tuning.4. Analysis Of Evaluation... · 4.2. Phase 2: Refined... · 5. Dense Map Filtering
  31. [31]
    Illustration of Gaussian process classification (GPC) on the XOR ...
    This example illustrates GPC on XOR data. Compared are a stationary, isotropic kernel (RBF) and a non-stationary kernel (DotProduct).
  32. [32]
    [2402.06184] The boundary of neural network trainability is fractal
    Feb 9, 2024 · The boundary between neural network hyperparameters that lead to stable and divergent training is fractal, over more than ten decades of scale.
  33. [33]
    Understanding Black-Box Models by Decision Boundary - MDPI
    Nov 5, 2024 · In this study, we provide notions of interpretability based on approximation theory. We first define explainable models in terms of explicitness and then use ...
  34. [34]
    [PDF] NEURAL NETWORK DECISION BOUNDARY VOLUME AND ITS ...
    linear support vector machine, for which the decision boundary is a co-dimension one hyperplane. The structure of the decision boundary on the entire ...
  35. [35]
    [PDF] smoothing decision boundaries to avoid overfitting in neural network ...
    Abstract: This work addresses the problem of overfitting the training data. We suggest smoothing the decision boundaries by eliminating border instances ...
  36. [36]
    [PDF] Characterizing the Decision Boundary of Deep Neural Networks
    First, we measure the complexity of the decision boundary in the input space. To this end, we measure the classification oscillation along the decision boundary ...
  37. [37]
    Exploring and Exploiting Decision Boundary Dynamics for ... - arXiv
    Feb 6, 2023 · Abstract:The robustness of a deep classifier can be characterized by its margins: the decision boundary's distances to natural data points.
  38. [38]
    [PDF] Boundary thickness and robustness in learning models
    We show that thick decision boundaries have a regularization effect that improves robustness, while thin decision boundaries lead to overfitting and reduced ...
  39. [39]
    What is the role of decision boundaries in Explainable AI? - Milvus
    In classification tasks, a decision boundary represents the threshold at which the model switches its prediction from one class to another.
  40. [40]
    LIME - Local Interpretable Model-Agnostic Explanations
    Apr 2, 2016 · In this post, we'll talk about the method for explaining the predictions of any classifier described in this paper, and implemented in this open source package.
  41. [41]
    Explainable AI, LIME & SHAP for Model Interpretability - DataCamp
    May 10, 2023 · They focus on analyzing the features' input-output pair. This section will introduce and discuss LIME and SHAP, two widely-used surrogate ...
  42. [42]
    Deciphering Model Decisions: A Comparative Analysis of SHAP and ...
    Mar 24, 2024 · SHAP and LIME offer valuable insights into the model's decision-making process from different perspectives.
  43. [43]
    make_moons — scikit-learn 1.7.2 documentation
    Make two interleaving half circles. A simple toy dataset to visualize clustering and classification algorithms. Read more in the User Guide. Parameters:.Missing: paper | Show results with:paper
  44. [44]
    Comparing Classification Algorithms using the Iris Dataset (Logistic ...
    Aug 31, 2025 · Decision tree's performance. Step 9: Comparing Results. Logistic Regression achieved an overall accuracy of 93.3%. It classified all setosa ...Missing: synthetic half- moon
  45. [45]
    Solving Nonlinearly Separable Classifications in a Single-Layer ...
    Mar 1, 2017 · Our primary focus was the XOR problem central to the critique made by Minsky and Papert (1969). ... Notably, the failure of a single-layer network ...
  46. [46]
    Understanding the Decision Boundary of Deep Neural Networks
    Feb 5, 2020 · We study the minimum distance of data points to the decision boundary and how this margin evolves over the training of a deep neural network.
  47. [47]
    [PDF] Topological Data Analysis of Decision Boundaries with Application ...
    Using the real-world application domains MNIST, FashionMNIST and CIFAR10, we show how these approaches can be used to evaluate the Decision Boundary ...
  48. [48]
    Enhancing breast cancer diagnosis through machine learning ...
    Jul 2, 2025 · Using kernels allows SVMs to create complex decision boundaries ... biomarkers for diagnosis, prognosis, and treatment decision-making.
  49. [49]
    Discovering biomarkers associated and predicting cardiovascular ...
    Jan 2, 2024 · We have uncovered 18 transcriptomic biomarkers that are highly significant in the CVD population that were used to predict disease with up to 96% accuracy.
  50. [50]
    Optimal learning of high-dimensional classification problems using ...
    Dec 23, 2021 · We study the problem of learning classification functions from noiseless training samples, under the assumption that the decision boundary is of a certain ...Missing: challenges | Show results with:challenges