15results about How to "Improve attack efficiency" patented technology

The invention relates to the technical field of information security, in particular to a deep learning side channel analysis method for efficiently recovering a key, which comprises the following steps: S1, acquiring plaintext, key and encrypted energy leakage information; s2, byte labels set by bytes corresponding to the secret key and energy traces in the interest interval correspondingly form a data set, and the data set is divided into a training set, a verification set and a test set; s3, constructing a single convolutional neural network model according to the relationship between the energy leakage generated in the interest interval and the key, and establishing a side channel analysis deep learning model; s4, inputting the training set into a side channel analysis deep learning model for training to obtain a final model, and collecting training indexes; and S5, evaluating the final model, recovering the byte key by using the final model, and evaluating the attack effect. According to the method, the key of n bytes can be recovered only by training one model, so that the number of models is reduced, the training time is saved, and the efficiency of deep learning side channel analysis is effectively improved.

The invention provides a multi-feature fusion-based black box attack detection method (MFBA), device and system, different features are extracted by using a multi-model fusion strategy, the effectiveness and quality of opposite samples are improved, and the MFBA comprises the following three key steps: firstly, obtaining an initial data set from a target model; a decision boundary is approached by using a multi-feature fusion model; secondly, the stability of the model is enhanced through multiple rounds of training application data expansion and incremental learning technologies; and finally, generating opposite examples by using the optimized fusion model. By constructing the comprehensive attack framework, how to combine various model features to improve the performance against the resistance attack is demonstrated; compared with a single substitution model, the antagonism examples generated by the MFBA have higher mobility under different disturbance levels and have better attack effects; according to the method, an effective solution is provided for black box antagonism attacks, and the model in the generation of the advantage opposite paradigm of multi-feature fusion is verified.

The present invention discloses an invalid sample optimization method and device for a boundary black box attack on a video model. The method comprises: firstly establishing current stage parameters according to the number of visits to the video model by the current iteration of the boundary black box attack; then generating a scaling gradient based on the current stage parameters; and finally updating the sample data of the previous iteration by the scaling gradient, which can effectively reduce the number of invalid samples in each iteration and greatly improve the efficiency of the boundary black box attack.

The application relates to the field of power systems and artificial intelligence security, and discloses a simulation attack and defense method of a power grid load frequency control system adopting a DRL controller, which comprises the following steps of: simulation attack: identifying system state variables most critical to DRL decision; setting an attack target and calculating an initial disturbance amount of the critical state variables, limiting the generation of a final disturbance amount; deducing disturbance amounts of the remaining dependent state variables to obtain an adversarial disturbance vector, which is superimposed to the DRL controller to implement the attack. Simulation defense: applied to a hybrid system containing DRL and a backup PID controller: real-time monitoring of the Q value sequence output by the DRL evaluation network to detect whether it is continuously lower than the normal benchmark; once the statistical quantity exceeds the threshold, it is determined that an attack is suffered, and the control right is immediately switched from the DRL to the PID controller to maintain the stability of the system. The method can generate a hidden and physically regular adversarial attack to evaluate the vulnerability of the system, and provide a fast and low-cost active defense mechanism to improve the security and resilience of the LFC system in the adversarial environment.

The invention discloses an adversarial sample generation method and system, a storage medium and a terminal, and belongs to the technical field of adversarial attacks, and the method comprises the steps: determining a context region of a target object, and dividing the context region into a plurality of sub-regions; carrying out importance analysis on each sub-region; mapping the influence degree of each sub-region on the detection result of the target detection model to a context region, and generating a context attribution graph; selecting one or more influence sub-regions exceeding an influence threshold; performing fine tuning on the influence sub-region under a total loss function (including a loss function and a perception loss function) to generate an adversarial sample; according to the method, the attack direction is accurately guided through the context attribution graph, and the double loss functions are introduced, so that the imperceptibility of confrontation disturbance in vision is ensured while the position, direction and confidence coefficient prediction are inhibited, the integrity of a target object is ensured, and the attack concealment is enhanced. And attack is carried out only aiming at partial influence sub-regions, so that the attack efficiency is improved.

The invention discloses a virtual character control method and device, a storage medium and electronic equipment. The method comprises the steps of displaying a first virtual scene, entering a second virtual scene from the first virtual scene in response to a target virtual character, starting a target task, displaying a first virtual object in the second virtual scene, and displaying a target virtual object in the second virtual scene in response to the fact that the first virtual object is corrupted, and in response to the situation that the attack suffered by the target virtual object within the preset duration does not meet the preset condition, canceling display of the target virtual object, displaying a second virtual object in the second virtual scene, and in response to the situation that the virtual object to be attacked indicated by the target task does not exist in the second virtual scene, completing the target task. According to the method and the device, the technical problem that the control efficiency of the virtual role is relatively low due to the fact that more selection possibilities are not provided for the user after the specific virtual object set for the specific virtual scene is damaged in the related technology is solved.

The present invention discloses a cross-modal collaborative adversarial attack method and device, which relates to the field of artificial intelligence security. It is used to solve the current problems of difficulty in modeling cross-modal collaborative attacks and difficulty in building optimization mechanisms in the field of large language model adversarial attacks. The cross-modal collaborative attack provided by this method can circumvent these security alignment mechanisms based on single-modal content feedback, thereby breaking through existing protection measures and increasing the complexity and unpredictability of attacks; furthermore, by optimizing adversarial perturbations in key cross-modal semantic subspaces and combining historical and future gradient information, the efficiency, convergence speed and stability of perturbation optimization can be improved, avoiding falling into local optimal solutions, thereby improving the practicality and operability of attacks.

This invention provides an scalable AI attack benchmark testing method and system. A model conversion module transforms the tested model according to its type and framework, obtaining a converted model that matches the attack framework. The dataset attacks the converted model, yielding a test result. The attack framework invokes various attack algorithms with a standardized format, generates corresponding attack scripts, and then performs several attacks on the converted model, resulting in multiple attack results. When testing the next tested model, the model conversion module again transforms its framework according to its type before launching the attack. The advantages are: it can flexibly convert for different AI models, attacking the converted model to obtain attack results, without needing to rebuild test modules for different AI models, improving attack efficiency, reducing testing costs, and solving the problem of existing testing methods (frameworks) being limited by the tested model and language.

This invention discloses an adversarial example generation method, system, storage medium, and terminal, belonging to the field of adversarial attack technology. The method includes: determining the context region of a target object and dividing it into multiple sub-regions; performing importance analysis on each sub-region; mapping the influence of each sub-region on the detection results of the target detection model to the context region, generating a context attribution graph; selecting one or more influential sub-regions exceeding an influence threshold; and fine-tuning the influential sub-regions under a total loss function (including a loss function and a perceptual loss function) to generate adversarial examples. This invention accurately guides the attack direction through the context attribution graph and introduces a dual loss function, suppressing position, direction, and confidence prediction while ensuring the visual imperceptibility of adversarial perturbations, thus preserving the integrity of the target object and enhancing the stealth of the attack. Attacking only some influential sub-regions improves attack efficiency.

The invention provides a federated learning backdoor attack method based on backdoor structure disassembly, and the method comprises the steps: enabling an attacker to determine the distribution of a backdoor structure according to a misclassification target; the attackers turn off nerve cell node gradient switches except key nerve cell set Wk nodes in the global model Gt, training the global model Gt through a local poisoning sample # imgabs0 # to generate a local poisoning model # imgabs1 #, and after the local poisoning model # imgabs2 # is obtained, splitting a weight set # imgabs3 # corresponding to a back door structure in the local poisoning model # imgabs2 # and distributing the weight set # imgabs3 # to each attacker; the attacker i adjusts the neuron weight attackeri corresponding to the key neuron set Wk; each participant uploads the gradient parameters to a central server; s6, model aggregation stage: aggregating gradient parameters uploaded by each participant by a central server, generating a new global model Gt by combining the global model Gt with the parameters uploaded by each participant, replacing a part corresponding to a key neuron set Wk of the aggregated global model with a weight set # imgabs4 # corresponding to a backdoor structure, and replacing the global model Gt with a local poisoning model # imgabs5 #

The present application discloses a method and device for generating adversarial samples, which are applied to the field of adversarial attack technology. The method includes: acquiring and processing user images to obtain a first image set and an adversarial sample set; determining frequency domain perturbations and fine-tuning a diffusion model according to the first image set to obtain a first diffusion model and a heat map set; updating the adversarial sample set N times in the frequency domain to obtain a first sample set, updating the first sample set M times in the spatial domain to obtain a second sample set; updating the first diffusion model, updating the second sample set M times in the spatial domain to obtain a third sample set, updating the third sample set N times in the frequency domain to obtain a fourth sample set as an adversarial sample set, iterating the above steps multiple times, thereby obtaining an adversarial sample set, wherein the frequency domain update is realized by the first diffusion model, the frequency domain perturbations and the heat map set, and the spatial domain update is realized by the first diffusion model and the heat map set. The present invention improves the effect and efficiency of adversarial attacks and ensures the image quality of adversarial samples.

The embodiment of the invention discloses a method and device for evaluating robustness of a diffusion-based purification model. The specific implementation mode of the method comprises the steps that random purification is carried out on a noise image for a first number of times through a purification model, a first number of purification results are obtained, and the noise image is obtained after adversarial noise is superimposed on a sample image; determining a first number of classification results according to the first number of purification results through a classifier; determining a loss value and a gradient in each random purification process based on the difference between each classification result and the actual category of the sample image; determining a representative random purification process based on the loss value and gradient in each random purification process; performing reverse gradient propagation based on a representative random purification process, and determining a target gradient; and attacking the purification model based on the target gradient, and determining the robustness of the purification model. According to the embodiment, the accuracy of evaluating the robustness of the diffusion-based purification model is improved.

The application provides a heat map guidance-based image quality evaluation model attack method and system, and belongs to the technical field of black-box attack. The method comprises the following steps: S1, obtaining an original image and a target image quality evaluation model, setting a perturbation constraint and a query parameter, generating an initial adversarial sample and initializing a sensitivity heat map; S2, selecting an image perturbation region according to the sensitivity heat map, updating the perturbation in combination with a dynamic scheduling strategy, generating a candidate adversarial sample and obtaining a model output; S3, performing reward, punishment and smoothing update on the sensitivity heat map according to the target function change of the candidate sample and the current sample; and S4, outputting a final adversarial sample when a termination condition is reached. The application can effectively solve the problems of low attack success rate, poor query efficiency and insufficient generalization ability of the existing adversarial attack method in the image quality evaluation task, and provides technical support for the robustness test, security evaluation and defense mechanism design of the image quality evaluation model.

The invention discloses a cross-modal collaborative countermeasure attack method and device, and relates to the field of artificial intelligence security. The method and the device are used for solving the problems of difficulty in cross-modal cooperative attack modeling and difficulty in optimization mechanism construction in the field of large language model countermeasure attacks at present. The cross-modal cooperative attack provided by the method can avoid security alignment mechanisms based on single-modal content feedback, so that the existing protection means is broken through, and the complexity and unpredictability of the attack are improved; moreover, by optimizing adversarial disturbance in the key cross-modal semantic subspace and combining historical and future gradient information, the efficiency, convergence speed and stability of disturbance optimization can be improved, and falling into a local optimal solution is avoided, so that the practicability and operability of attacks are improved.

The invention provides a collision attack post-processing method and system based on a genetic algorithm. The method comprises the following steps: firstly, acquiring multiple groups of random plaintexts, and encrypting through a cryptographic algorithm to obtain corresponding power consumption traces; performing correlation enhanced collision attack on the random plaintext and the power consumption trace to obtain a possibility ranking of collision values of every two key bytes; calculating an energy trace average value of each group of power consumption traces, and calculating a difference sum of squares of the energy traces among the key bytes; according to the theoretical distribution of the sum of squares of the difference values, confirming a collision value between every two key bytes; the collision values of every two secret key bytes are rearranged, the values marked as collision are moved forwards, and the values marked as non-collision are moved backwards; and inputting the initial population into a genetic algorithm framework for evolution, performing selection, crossover and mutation operations until an evolution algebra reaches a threshold value or a correct collision chain is found, and outputting an individual with the highest fitness as a result. By optimizing the evaluation and selection strategy of the key candidates, the efficiency and accuracy of collision attacks are remarkably improved.