Statistical result dynamic visualization and diagnosis method based on explainable ai

Abstract

The application relates to the technical field of artificial intelligence, and discloses a statistical result dynamic visualization and diagnosis method based on explainable AI, which comprises the following steps: for each data sample of a machine learning model, a plurality of explainable algorithms are applied in parallel to generate and splice into high-dimensional integrated explanation vectors; the similarity between vectors is calculated through a plurality of component explanation kernel functions, and a manifold learning algorithm is used to reduce the high-dimensional vector set, so that a low-dimensional explanation manifold reflecting the internal structure of model decision logic is constructed; through density clustering of the explanation manifold, explanation stable domains with consistent model decision logic and logic mutation boundaries with logic mutation are automatically identified; the analysis result is visually presented, and an interactive diagnosis function is provided. The application can make users intuitively understand and diagnose the behavior mode of a complex model by constructing a global topological view of model decision logic, and can significantly improve the explainability, transparency and reliability of the model.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a method for dynamic visualization and diagnosis of statistical results based on interpretable AI. Background Technology

[0002] In recent years, machine learning models, represented by deep neural networks and ensemble models, have achieved remarkable results in many fields such as image recognition, natural language processing, and financial risk control. However, the powerful performance of these models often comes at the cost of the high complexity and opacity of their internal decision-making logic, making them often referred to as "black boxes." This "black box" characteristic greatly limits the application of these models in high-risk and high-reliability scenarios such as healthcare, autonomous driving, and the judiciary, because users not only need to know the model's predictions, but also need to understand and trust the reasons behind those predictions.

[0003] To open the "black box," interpretable artificial intelligence (XAI) technology has emerged. In existing technologies, methods such as SHAP (SHapley Additive exPlanations) or LIME (Local Interpretable Model-agnostic Exlanations) are widely used to generate local explanations for a single data sample, that is, to clarify how the model makes a specific prediction for a particular input.

[0004] While these local interpretation methods improve model transparency to some extent, their application still has deep-seated limitations. These methods essentially provide a microscopic perspective, like "seeing the whole picture through a tube." To understand the model's global behavioral patterns, users must examine numerous isolated, discrete local interpretations and attempt inefficient and subjective induction and reconstruction in their minds. This process is not only cumbersome but also fails to reveal the macroscopic structure, internal connections, and transitional relationships between different regions of the model's decision-making logic within the entire data space.

[0005] Furthermore, existing technologies have failed to effectively address the issue of interpretability stability. The decision-making logic of a model is not static; even minor changes in input data can lead to drastic, non-linear "abrupt changes" in its interpretation. Existing methods struggle to systematically identify regions where the model's decision-making logic remains stable, nor can they clearly define the boundaries where logical abrupt changes occur. Users therefore cannot predict under what circumstances the model's "thinking style" will fundamentally shift, directly impacting their ability to predict and trust the model's behavior. Moreover, current technological paradigms mostly provide static, one-off analysis reports, lacking the ability to dynamically interact with users. Users cannot actively probe and diagnose the model's logical boundaries by adjusting inputs in real time, making it difficult to conduct deeper exploration and verification of the model's internal mechanisms. Summary of the Invention

[0006] Existing interpretable AI methods typically generate static, isolated explanations. This makes it difficult for users to dynamically and interactively diagnose the decision-making logic of machine learning models, and fails to systematically reveal the stability of the model's inherent decision-making patterns and the transition boundaries between different decision-making patterns.

[0007] The present invention aims to solve the above-mentioned technical problems and provide a method for dynamically visualizing and diagnosing the statistical results of AI models, so as to realize the automated analysis and display of model decision patterns, stable regions and logical mutation boundaries.

[0008] The first aspect of this invention provides a method for dynamic visualization and diagnosis of statistical results based on interpretable AI, comprising the following steps:

[0009] Obtain multiple data samples for a predefined machine learning model;

[0010] For each of the plurality of data samples, a high-dimensional integrated interpretation vector is generated by applying at least two interpretability AI algorithms and concatenating their interpretation results.

[0011] Based on the multiple ensemble explanatory vectors corresponding to the multiple data samples, a low-dimensional explanatory manifold that displays the relationship between the multiple ensemble explanatory vectors is constructed through a manifold learning algorithm;

[0012] A topological analysis is performed on the explanatory manifold to identify explanatory stable regions representing stable decision-making patterns, as well as logical abrupt boundary boundaries between the explanatory stable regions;

[0013] The explanatory manifold, the explanatory stable region, and the logical mutation boundary are visualized.

[0014] In one specific embodiment, the at least two interpretable AI algorithms include two or more of the following: game theory-based methods, local surrogate model-based methods, or gradient-based methods. The interpretable component vectors generated in this step are concatenated to form the integrated interpretable vector.

[0015] In one specific embodiment, the process of determining the relationship between the multiple integrated interpretation vectors in the step of constructing the low-dimensional interpretive manifold is achieved by calculating the value of a multi-component interpretive kernel function. The step of calculating the multi-component interpretive kernel function value specifically includes:

[0016] For any two ensemble interpretation vectors and Calculate the values ​​of at least two of the following sub-kernel functions and perform a weighted sum:

[0017] 1. Orientation similarity kernel function Its value is used to characterize the directional consistency of the two integrated interpretation vectors. A specific calculation method is:

[0018] ;

[0019] in, For vectors and The Euclidean distance between them This is the preset bandwidth parameter.

[0020] 2. Order Consistency Kernel Function Its value is used to characterize the similarity of the feature contribution ranking in the two integrated explanatory vectors. A specific calculation method is to calculate the Spearman rank correlation coefficient of the element values ​​in the two vectors.

[0021] 3. Key order of magnitude kernel function Its value is used to characterize the similarity of the contribution levels of a predetermined number of key features in the two integrated explanation vectors. A specific calculation method is:

[0022] ;

[0023] in, It is an extraction vector The highest absolute value A function of a subvector composed of n elements The number of key features is preset.

[0024] Preferably, the manifold learning algorithm includes at least one of UMAP, Isomap, or t-SNE.

[0025] In one specific embodiment, the step of identifying and interpreting the stable region specifically includes:

[0026] In the low-dimensional space where the explanatory manifold is located, a density clustering algorithm is applied to identify the points in the explanatory manifold as at least one high-density cluster and boundary points; wherein, the high-density cluster is defined as the explanatory stable region.

[0027] Furthermore, the method further includes calculating the boundary steepness of the logical mutation boundary. The boundary steepness... Defined as:

[0028] The distance between two points belonging to different interpretive stability domains on either side of the logical mutation boundary in the original high-dimensional integrated interpretive vector space. The distance between the two points in the low-dimensional interpretive manifold space The ratio. The specific calculation method is as follows:

[0029] ;

[0030] in, The kernel function can be interpreted through the aforementioned multi-component approach. Export the value. Let be the Euclidean distance between two points in a low-dimensional space.

[0031] Preferably, the visualization step further includes:

[0032] The numerical value of the boundary steepness is displayed on the logically abrupt boundary using visual attributes such as color, thickness, or transparency.

[0033] In one specific embodiment, the method further includes:

[0034] In response to the selection operation of any of the aforementioned interpretive stable regions, the feature statistics and average ensemble interpretive vector of all data samples within the interpretive stable region are aggregated to generate a semantic description of the interpretive stable region.

[0035] In one specific embodiment, the method further includes:

[0036] In response to a modification to the original feature value of any data sample, the integrated interpretation vector of the modified data sample is recalculated in real time.

[0037] Based on the updated integrated interpretation vector, determine its new position on the interpretation manifold;

[0038] The visualization presents the trajectory of the data sample as it moves from an old position to a new position on the explanatory manifold.

[0039] A second aspect of the present invention provides a dynamic visualization and diagnostic system for statistical results based on interpretable AI, comprising:

[0040] processor;

[0041] A memory that stores computer-executable instructions;

[0042] When the processor executes the computer-executable instructions, it is configured to perform the method described in the first aspect above.

[0043] This invention provides a method for dynamic visualization and diagnosis of statistical results based on interpretable AI. It has the following beneficial effects:

[0044] 1. This invention can systematically reveal the inherent decision-making patterns of machine learning models. First, it generates an ensemble explanatory vector for each data sample. Then, it constructs an explanatory manifold to represent the relationships between these vectors and further automatically identifies the explanatory stability region and logistic mutation boundary through topological analysis. This process transforms a series of discrete, local explanatory results into a structured, global view of decision logic, thereby enabling a systematic analysis of the model's behavioral patterns across the entire dataset.

[0045] 2. This invention can quantitatively evaluate the stability of the model's decision logic. This is achieved by calculating the boundary steepness after identifying the logical abrupt change boundaries between different explanatory stable domains. The boundary steepness is defined as the ratio of the distance change in the high-dimensional integrated explanatory vector space to the distance change in the low-dimensional explanatory manifold space. This ratio provides a specific and comparable value for the degree of drastic logical transition of the model at the decision boundary, thereby realizing a quantitative evaluation of the model's robustness.

[0046] 3. This invention provides an intuitive method for dynamically diagnosing model decision logic. When a user modifies the original feature value of any data sample, the system can recalculate its integrated explanatory vector and update its position on the explanatory manifold in real time. By presenting this positional change as a visual trajectory, users can directly observe how the data sample point moves between different explanatory stability regions and logistic abrupt change boundaries, thereby intuitively understanding the causal relationship between changes in input data and shifts in model decision logic. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the hardware structure of a statistical result dynamic visualization and diagnosis system based on interpretable AI according to an embodiment of the present invention;

[0048] Figure 2 This is a functional block diagram of a statistical result dynamic visualization and diagnostic system based on interpretable AI, according to an embodiment of the present invention.

[0049] Figure 3 This is a flowchart of a method for dynamic visualization and diagnosis of statistical results based on interpretable AI, according to an embodiment of the present invention.

[0050] Figure 4 This is a schematic diagram illustrating the manifold construction process according to an embodiment of the present invention.

[0051] Among them, 101 is the processor; 102 is the memory; 103 is the system bus; 104 is the input / output interface; 105 is the communication interface; 10 is the data acquisition module; 20 is the integrated interpretation vector generation module; 30 is the interpretation manifold construction module; 40 is the topology analysis module; and 50 is the visualization and interaction module. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, specific embodiments of the invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0053] See attached document Figure 1 , Figure 1 This is a schematic diagram of the hardware structure of a dynamic visualization and diagnostic system for statistical results based on interpretable AI, according to an embodiment of the present invention. The system can be a server, personal computer, workstation, or any device with computing capabilities. The system includes: at least one processor 101, such as a central processing unit (CPU); a memory 102 for storing program code and data; a system bus 103 for connecting the processor 101 and the memory 102, as well as other components; an input / output interface 104 for connecting external devices such as a monitor, keyboard, and mouse; and a communication interface 105 for data communication with other devices. The memory 102 stores computer program instructions, which, when executed by the processor 101, implement the methods described later.

[0054] See attached document Figure 2 , Figure 2 This is a functional block diagram of a dynamic visualization and diagnostic system for statistical results based on interpretable AI, according to an embodiment of the present invention. In this embodiment, the system may include: a data acquisition module 10, an integrated interpretation vector generation module 20, an interpretation manifold construction module 30, a topology analysis module 40, and a visualization and interaction module 50.

[0055] The data acquisition module 10 is used to acquire a preset, pre-trained machine learning model from the memory 102 or from an external data source via the communication interface 105, and to acquire multiple data samples for diagnosing the model. After acquisition, the machine learning model and multiple data samples are provided to the integrated interpretation vector generation module 20.

[0056] The integrated interpretable vector generation module 20 receives machine learning models and data samples provided by the data acquisition module 10. For each data sample, this module calls at least two different types of interpretability AI algorithm libraries to calculate the corresponding interpretable component vectors, and then concatenates the interpretable component vectors in a predetermined order to form a high-dimensional integrated interpretable vector. This module outputs the set of integrated interpretable vectors corresponding to all data samples to the interpretable manifold construction module 30.

[0057] The interpretive manifold construction module 30 receives the set of ensemble interpretive vectors output by the ensemble interpretive vector generation module 20. This module first calculates the similarity value between any two ensemble interpretive vectors based on a preset multi-component interpretive kernel function, thereby constructing a similarity matrix. Subsequently, the module uses the similarity matrix as input to a preset manifold learning algorithm for dimensionality reduction calculation, ultimately generating a set of low-dimensional coordinate points that represent the relationships between the ensemble interpretive vectors. This low-dimensional coordinate point set is defined as the interpretive manifold and is output to the topology analysis module 40 and the visualization and interaction module 50.

[0058] The topology analysis module 40 receives the interpretive manifold output by the interpretive manifold construction module 30. This module first performs a density clustering algorithm on the low-dimensional coordinate point set of the interpretive manifold to identify high-density point clusters and boundary points, where high-density point clusters are defined as interpretive stable regions and boundary points are defined as logistic abrupt boundaries. Subsequently, the module calculates the boundary steepness at each point on the logistic abrupt boundaries. This module outputs the identified interpretive stable regions, logistic abrupt boundaries, and calculated boundary steepness data to the visualization and interaction module 50.

[0059] The visualization and interaction module 50 receives interpretive manifold data from the interpretive manifold construction module 30 and analysis result data from the topology analysis module 40. This module renders this data into a graphical user interface and displays it on a display device. Simultaneously, this module also receives user operation commands transmitted via an input device and executes corresponding interactive functions based on the commands. For example, when receiving a user command to modify the original feature values ​​of a data sample, this module sends the modified data sample to the integrated interpretive vector generation module 20 for triggered recalculation.

[0060] See attached document Figure 3 , Figure 3 This is a flowchart of a method for dynamic visualization and diagnosis of statistical results based on interpretable AI, according to an embodiment of the present invention. The specific steps of this method may include:

[0061] Step S301: Obtain a pre-set, trained machine learning model and multiple data samples for diagnosing the machine learning model.

[0062] Step S302: For each data sample among multiple data samples, apply at least two interpretability AI algorithms to calculate the corresponding interpretability component vectors, and concatenate the interpretability component vectors to generate a high-dimensional integrated interpretability vector.

[0063] Step S303: Based on the integrated explanatory vector set corresponding to all data samples, calculate the pairwise similarity by applying a preset multi-component explanatory kernel function, and use the manifold learning algorithm to reduce dimensionality, thereby constructing a low-dimensional explanatory manifold.

[0064] Step S304: Perform topological analysis on the explanatory manifold, identify the explanatory stable region and the logical mutation boundary by applying the density clustering algorithm, and further calculate the boundary steepness of the logical mutation boundary.

[0065] Step S305: Integrate and render the interpretive manifold, as well as the interpretive stable regions and logical mutation boundaries marked on the interpretive manifold, and finally generate and display a visual diagnostic interface.

[0066] The following will provide a detailed explanation of each step in the above process.

[0067] exist Figure 3 In the illustrated process, step S301 is the preparatory work before performing the diagnosis. This step first requires obtaining a pre-trained machine learning model. The machine learning model can be any type of supervised learning model, such as a logistic regression model, support vector machine, gradient boosting decision tree (GBDT), deep neural network (DNN) for classification tasks, or a linear regression model, regression tree, etc. for regression tasks. The model's state is fixed, meaning its internal parameters, such as weights, biases, or tree structure, have been determined through the training process, and the model already has interface functions for predicting new data.

[0068] This step requires acquiring multiple data samples for diagnostics. These samples can originate from training sets used to train the model, validation sets used to tune hyperparameters, test sets used to evaluate model performance, or production datasets encountered in real-world applications. The structure and format of each data sample must match the input requirements of the machine learning model. For example, for a model processing tabular data, each data sample is a feature vector, with each dimension corresponding to a specific business or physical feature.

[0069] The acquired data sample set should be large enough to support subsequent topological analysis in order to identify statistically significant patterns. Furthermore, the data sample set should cover a variety of input scenarios the model might encounter, including samples correctly predicted by the model, samples incorrectly predicted by the model, and samples with low confidence levels in the model's predictions. This provides the necessary data foundation for a comprehensive analysis of the model's decision-making behavior.

[0070] exist Figure 3In the process shown, step S302 is to generate a high-dimensional ensemble interpretation vector for each data sample that can comprehensively describe the basis on which the machine learning model makes decisions.

[0071] For each data sample obtained from step S301 The system applies a pre-defined set of interpretable AI algorithms, consisting of at least two different principles, either in parallel or sequentially. The algorithm set may include:

[0072] Game theory-based algorithms, such as SHAP (Shapley Additive exPlanations), assign a contribution value to the prediction result for each feature by calculating the marginal contribution of each feature across all possible feature combinations and using the Shapley value as the result. Applying this algorithm yields an explanatory component vector. Its dimensions are the same as the feature dimensions of the data samples.

[0073] Algorithms based on local surrogate models, such as LIME (Local Interpretable Model-agnostic Explanations). This algorithm is used in data samples... A set of perturbation samples is generated within the neighborhood of the original sample, and these perturbation samples are predicted using a pre-defined machine learning model. Subsequently, a simple model with interpretability (such as a linear regression model or decision tree) is used to fit these perturbation samples and their predictions. The parameters of this simple model (such as the weights of a linear model) are then used as the basis for evaluating the original sample. The interpretation of the prediction results forms the interpretation component vector. .

[0074] When the pre-defined machine learning model is a differentiable model (such as a deep neural network), the algorithm set can also include gradient-based methods. These methods calculate the model's output value relative to the input data samples. The gradient vector is obtained by taking the partial derivatives of each feature dimension. This gradient vector, or its derivatives (e.g., the product of the gradient and the input), can be used as the explanatory component vector. This indicates the sensitivity of the model output to small changes in the input features.

[0075] After calculating the explanatory component vectors corresponding to all preset algorithms, this step concatenates these vectors in a fixed, predefined order to form a higher-dimensional, more comprehensive integrated explanatory vector. This splicing operation can be represented as:

[0076] ;

[0077] in, This represents the vector concatenation operation. By performing this step on all data samples, a final set of multiple ensemble interpretation vectors is generated. ,in This represents the total number of data samples. This set will be used for subsequent interpretative manifold construction.

[0078] See attached document Figure 4 , Figure 4 This is a schematic diagram of the explanatory manifold construction process according to an embodiment of the present invention. In step S303, this step receives the integrated explanatory vector set generated in step S302. This process maps the high-dimensional space to a low-dimensional space to construct an interpretable manifold. Specifically, this step involves two stages: calculating similarity values ​​and performing dimensionality-reduced embedding.

[0079] During the similarity value calculation stage, the system interprets the kernel function using a pre-defined multi-component approach. To measure any two ensemble interpretation vectors and The degree of similarity between them. The value of the kernel function is defined as a weighted sum of the values ​​of multiple sub-kernel functions, in the form of:

[0080] ;

[0081] in, , , These are preset weight coefficients that are greater than zero, and their sum is 1.

[0082] The first sub-kernel function is the orientation similarity kernel function. This is used to measure the overall directional consistency of two integrated interpretation vectors in the vector space. In this embodiment, it is calculated using Gaussian radial basis functions:

[0083] ;

[0084] in, Representing vectors and The L2 norm between them, i.e., the Euclidean distance; It is a preset bandwidth hyperparameter that is greater than zero.

[0085] The second sub-kernel function is the sorting consistency kernel function. This is used to measure the consistency of the order in which the values ​​of each element in two ensemble interpretation vectors are sorted. The calculation process is as follows: First, two ensemble interpretation vectors are generated respectively. and Corresponding sorting vector and Each element in the sorted vector is the rank of its corresponding element in the original vector. Then, the two sorted vectors are calculated. and The Spearman rank correlation coefficient, the value of which is... .

[0086] The third sub-kernel function is the critical order kernel function. This is used to measure the proximity of the highest absolute value key features in a predetermined number of two ensemble explanatory vectors at the contribution level. The calculation process is as follows: First, based on a predetermined integer... From vector and Extract the values ​​with the highest absolute values ​​from the samples. Subvector consisting of elements and Then, its kernel function value is calculated using the following formula:

[0087] ;

[0088] in, It represents the L2 norm between two key feature vectors.

[0089] After defining the multi-component interpretation kernel function, this step involves integrating the interpretation vector set. The kernel function is applied to each pair of vectors in the matrix to construct an N×N symmetric similarity matrix. .matrix elements in The value is equal to .

[0090] In this embodiment, the weighting coefficients in the multi-component interpretation kernel function are... , , and the bandwidth hyperparameter in the orientation similarity kernel function. The number of key features in the key magnitude kernel function Its specific value can be predetermined based on experience, or determined by experimental methods such as grid search on the validation dataset, with the goal of obtaining the optimal visualization effect.

[0091] During the dimensionality reduction embedding stage, the similarity matrix... This is used as input to the manifold learning algorithm. In this embodiment, the Uniform Manifold Approximation and Projection (UMAP) algorithm can be employed. The UMAP algorithm is based on the similarity matrix. The algorithm constructs a weighted graph structure in a high-dimensional space and then searches for an optimal low-dimensional (e.g., 2D or 3D) embedding representation that best preserves the topological structure of the original high-dimensional space. The output of the algorithm is a set of low-dimensional coordinates. Each coordinate point Corresponding to the original integrated interpretation vector This set of coordinate points in low-dimensional space together constitutes the interpretive manifold for visualization.

[0092] It should be understood that although this embodiment uses the UMAP algorithm as an example, any manifold learning algorithm that can accept a similarity matrix or distance matrix as input and perform nonlinear dimensionality reduction is applicable to this invention. Other feasible algorithms include, but are not limited to: Isomap, t-distributed random neighborhood embedding (t-SNE), or locally linear embedding (LLE).

[0093] In step S304, this step receives the low-dimensional interpreted manifold generated in step S303 and performs automated analysis of its topology.

[0094] This step will first explain the manifold, which is the set of low-dimensional coordinate points. As input, a preset density clustering algorithm is applied. In this embodiment, the algorithm used is HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise). The HDBSCAN algorithm processes the input set of data points, dividing them into several clusters and a set of noise points according to the density distribution of the points. This algorithm does not require pre-specifying the number of clusters and can identify clusters of arbitrary shapes.

[0095] After the HDBSCAN algorithm has finished executing, each data point Each data point is assigned a label indicating its cluster or whether it is marked as noise. This step defines the region comprised of all data points assigned to any cluster label as the interpretable stable region of the interpretable manifold. Each independent cluster corresponds to an independent interpretable stable region. The set of all points marked as noise by the algorithm is defined as the logistic abrupt boundary between interpretable stable regions.

[0096] Subsequently, this step performs a quantitative analysis of the identified logistic abrupt change boundaries, specifically calculating their kurtosis. The calculation process is as follows: For entities belonging to two different interpretative stable domains... and any pair of neighboring points and Its boundary steepness Calculated.

[0097] First, calculate the distance between the two points in the low-dimensional interpretable manifold space. In this embodiment, the distance is the Euclidean distance:

[0098] ;

[0099] Then, based on the index and Retrieve the original high-dimensional ensemble interpretation vector generated in step S302 without dimensionality reduction. and Calculate the distance between these two high-dimensional vectors in the original space. In this embodiment, the distance is also a Euclidean distance:

[0100] ;

[0101] Finally, the boundary steepness is calculated using the following formula. :

[0102] ;

[0103] In this embodiment, high-dimensional distance and low-dimensional distance The calculations are all illustrated using Euclidean distance (L2 norm) as an example. Depending on the specific distribution characteristics of the data, other distance metrics can also be used, such as Manhattan distance (L1 norm) or Minkowski distance.

[0104] This step repeats the calculation for all points on the logistic mutation boundary or multiple pairs of nearby points to obtain a set of boundary steepness values. These values, along with the results of interpreting the stable region and identifying the logistic mutation boundary, are output for subsequent visualization.

[0105] In step S305, the purpose of this step is to integrate all the analysis results generated in the previous steps, generate a graphical user interface, and realize interactive diagnostic functions with the user.

[0106] First, this step generates a static visualization interface by receiving the set of low-dimensional coordinate points of the interpretive manifold generated in step S303. These points are then rendered as a scatter plot in a two-dimensional or three-dimensional coordinate system, forming the basic view for interpreting the manifold.

[0107] Next, the interpretable stability region information identified in step S304 is received, and an independent visual identifier is assigned to each interpretable stability region, for example, by using different colors or symbols to mark points belonging to different stability regions.

[0108] Simultaneously, it receives the identification results of logical abrupt change boundaries and the calculated boundary steepness values. For points defined as logical abrupt change boundaries, the system uses a different unified visual identifier (e.g., gray) distinct from the interpreted stable domain for labeling.

[0109] Furthermore, based on the boundary steepness value of each boundary point or neighboring point pair, it is rendered into different visual attributes through a preset visual mapping function. For example, the boundary steepness value is linearly or logarithmically mapped to the color depth, thickness, or transparency of the boundary line, where a higher boundary steepness value corresponds to a darker, thicker, or more opaque visual appearance.

[0110] Secondly, this step implements two core interactive diagnostic functions.

[0111] The first interactive function is the generation of semantic descriptions of interpretable stability regions. When the system detects that a user has selected a specific interpretable stability region on the visual interface (e.g., by clicking with the mouse) through input / output interface 104, the system will perform the following operations:

[0112] First, retrieve the indices of all data sample points belonging to this interpretation's stability region;

[0113] Then, based on these indices, the corresponding subset of data samples is extracted from the original dataset;

[0114] Next, statistical analysis is performed on each feature dimension of the data sample subset to calculate its mean, variance, median, mode, and other statistical measures.

[0115] At the same time, the system also calculates the average vector of all integrated interpretation vectors within the stability region.

[0116] Finally, the system combines these statistical information and key features from the average explanatory vector into a structured text as a semantic description of the explanatory stability domain, and displays it in a preset area of ​​the interface.

[0117] The second interactive function is dynamic trajectory tracking of data samples. When the system detects that the user has selected a data sample point on the interface and modified one or more feature values ​​of the corresponding original data sample, the system will trigger a real-time recalculation process. This process will pass the modified new data sample back to step S302 and execute some calculations in S302 and S303 in sequence. That is, for only this modified sample, its integrated interpretation vector will be regenerated, and its new position coordinates on the low-dimensional interpretation manifold will be calculated using the constructed manifold model.

[0118] After obtaining the new location coordinates, the visualization interface executes an animation process that smoothly moves the point representing the data sample from its old location to the new location, leaving a visual trajectory line along its movement path. This trajectory line visually demonstrates how changes in the input data cause the model's interpretation to shift its location on the interpretive manifold.

[0119] It is worth noting that when calculating the new position coordinates of the modified data samples, this embodiment utilizes the transform function of the manifold learning model (e.g., the trained UMAP model) that has been built and solidified in step S303. This process does not require retraining the manifold model for all data points, but only projects the new high-dimensional ensemble interpretation vector onto the existing low-dimensional interpretation manifold space. Therefore, the computational overhead is small, and it can support real-time or near real-time interactive feedback.

[0120] In summary, the method and system provided by the embodiments of the present invention generate integrated explanatory vectors for data samples, construct a low-dimensional explanatory manifold based on the vector set, analyze the topological structure of the manifold to identify explanatory stable regions and logical mutation boundaries, and finally visualize and interactively diagnose the analysis results, thereby realizing a systematic analysis of the internal decision-making logic of machine learning models.

[0121] Finally, it should be noted that the above descriptions are merely specific embodiments of the present invention, intended to illustrate the technical solutions of the present invention rather than to limit them. For those skilled in the art, any modifications to the technical solutions in the above embodiments, equivalent substitutions of some technical features, or combinations based thereon, made without departing from the spirit and principles of the present invention, should fall within the protection scope of the appended claims.

Claims

1. A method for dynamic visualization and diagnosis of statistical results based on interpretable AI, characterized in that, Includes the following steps: Obtain multiple data samples for a predefined machine learning model; For each of the plurality of data samples, a high-dimensional integrated interpretation vector is generated by applying at least two interpretability AI algorithms and concatenating their interpretation results. Based on the multiple ensemble explanatory vectors corresponding to the multiple data samples, a low-dimensional explanatory manifold that displays the relationship between the multiple ensemble explanatory vectors is constructed through a manifold learning algorithm; A topological analysis is performed on the explanatory manifold to identify explanatory stable regions representing stable decision-making patterns, as well as logical abrupt boundary boundaries between the explanatory stable regions; The explanatory manifold, the explanatory stable region, and the logical mutation boundary are visualized. The method further includes: Calculate the boundary steepness of the logical mutation boundary; The boundary steepness is defined as the ratio of the distance between two points belonging to different interpretive stable domains on both sides of the logical mutation boundary in the original high-dimensional integrated interpretive vector space to the distance between the two points in the low-dimensional interpretive manifold space. The visualization process also includes: The numerical value of the boundary steepness is displayed on the logically abrupt boundary using visual attributes such as color, thickness, or transparency. The method further includes: In response to a modification to the original feature value of any data sample, the integrated interpretation vector of the modified data sample is recalculated in real time. Based on the updated integrated interpretation vector, determine its new position on the interpretation manifold; The visualization presents the trajectory of the data sample as it moves from an old position to a new position on the explanatory manifold.

2. The method for dynamic visualization and diagnosis of statistical results based on interpretable AI according to claim 1, characterized in that, The steps of applying at least two interpretable AI algorithms specifically include: The application includes at least two of the following methods: game theory-based methods, local proxy model-based methods, or gradient-based methods, to generate corresponding explanatory component vectors, and then concatenates the explanatory component vectors into the integrated explanatory vector.

3. The method for dynamic visualization and diagnosis of statistical results based on interpretable AI according to claim 1, characterized in that, In the step of constructing a low-dimensional interpretive manifold that demonstrates the relationship between the multiple integrated interpretive vectors, the process of determining the relationship between the multiple integrated interpretive vectors is achieved by calculating the multi-component interpretive kernel function value; The steps for calculating the multi-component interpretation kernel function value specifically include: For any two of the aforementioned ensemble interpretation vectors, calculate the values ​​of at least two of the following sub-kernel functions and perform a weighted sum: The orientation similarity kernel function, whose value is used to characterize the orientational consistency of the two integrated interpretation vectors; The ranking consistency kernel function is used to characterize the similarity of the feature contribution ranking in the two integrated explanatory vectors; A key magnitude kernel function, the value of which is used to characterize the proximity of the contribution magnitudes of a predetermined number of key features in the two integrated interpretation vectors.

4. The method for dynamic visualization and diagnosis of statistical results based on interpretable AI according to claim 1, characterized in that, The manifold learning algorithm includes at least one of UMAP, Isomap, or t-SNE.

5. The method for dynamic visualization and diagnosis of statistical results based on interpretable AI according to claim 1, characterized in that, The specific steps for identifying the explanatory stability region representing the stable decision-making pattern are as follows: In the low-dimensional space where the explanatory manifold is located, a density clustering algorithm is applied to identify the points in the explanatory manifold as at least one high-density cluster and boundary points; wherein, the high-density cluster is defined as the explanatory stable region.

6. The method for dynamic visualization and diagnosis of statistical results based on interpretable AI according to claim 5, characterized in that, The method further includes: In response to the selection operation of any of the aforementioned interpretive stable regions, the feature statistics and average ensemble interpretive vector of all data samples within the interpretive stable region are aggregated to generate a semantic description of the interpretive stable region.

7. A dynamic visualization and diagnostic system for statistical results based on interpretable AI, characterized in that: include: processor; A memory that stores computer-executable instructions; When the processor executes the computer-executable instructions, it implements the method as described in any one of claims 1-6.

Images