Abstract: There are provided systems and methods of bot detection through explainable deep learning and rule violation codebooks from generative AI. A service provider, such as an electronic transaction processor for digital transactions, may provide computing services to users for processing various requests and interactions with those users. However, malicious users may utilize bots, such as automated scripts and software applications, that attempt to conduct fraud, compromise systems and data, and the like. To provide better bot and bot activity detection, the service provider may implement an explainable deep learning system that may generate rule violations of rules indicating bot activity or presence in computing logs and interactions using a generative AI. The violations may have a corresponding explanation in codebooks to automate bot detection. When bot activity is detected, the explanation may provide a reason for the bot activity detection.

Abstract: There are provided systems and methods of bot detection through explainable deep learning and rule violation codebooks from generative AI. A service provider, such as an electronic transaction processor for digital transactions, may provide computing services to users for processing various requests and interactions with those users. However, malicious users may utilize bots, such as automated scripts and software applications, that attempt to conduct fraud, compromise systems and data, and the like. To provide better bot and bot activity detection, the service provider may implement an explainable deep learning system that may generate rule violations of rules indicating bot activity or presence in computing logs and interactions using a generative AI. The violations may have a corresponding explanation in codebooks to automate bot detection. When bot activity is detected, the explanation may provide a reason for the bot activity detection.

Abstract: Techniques are disclosed relating to determining whether input data is authentic. A system detects input data, that includes text data and typing data, at a computing device. The system may generate, using a string model, a string-level prediction for the input data, where the string model is trained to increase a similarity between embeddings of authentic text data and corresponding sequences of typing data. Using a character model, the system may generate a character-level prediction for the set of input data, where the character-level model predicts an intended sequence of characters based on the text data and a sequence of typing actions included in the input data. Using machine learning, the system determines, based on the string-level prediction and the character-level prediction, whether the input data is authentic input. The system transmits, to the device, a decision that is generated based on determining whether the input data is authentic.

Abstract: A method includes partitioning a set of historical device keys into a set of clusters, embedding the set of clusters to generate a set of device lockers, and receiving a request including an application identifier and metadata. The application identifier corresponds to an application installed on the second computing system and the metadata corresponds to the second computing system. The method also includes combining the application identifier and the metadata to generate a device key, determining that the set of device lockers includes a device locker that matches the device key, and, in response to determining that the set of device lockers includes the device locker that matches the device key, adding the device key to a cluster of the set of clusters corresponding to the device locker.

Abstract: Techniques for predicting whether a submission includes a forged image. A computer system receives a submission from a user that includes an image and image metadata, such as an identifier for the user and a User-Agent string value. An image pixel embedding is generated from the image, and a profile embedding is generated from the image metadata. The image embedding is indicative of whether the image is similar to known image forgeries. The profile embedding is generated from a user activity embedding indicative of User-Agent values associated with the user identifier. The profile embedding is generated using a machine learning model that uses stored parameters to associate user activity, device information, and forgery groups. The profile embedding thus indicates whether the user is associated with known image forgeries. The image pixel embedding and profile embedding are then used by a neural network to output a forgery prediction.

Abstract: Techniques are disclosed relating to determining whether input data is authentic. A system detects input data, that includes text data and typing data, at a computing device. The system may generate, using a string model, a string-level prediction for the input data, where the string model is trained to increase a similarity between embeddings of authentic text data and corresponding sequences of typing data. Using a character model, the system may generate a character-level prediction for the set of input data, where the character-level model predicts an intended sequence of characters based on the text data and a sequence of typing actions included in the input data. Using machine learning, the system determines, based on the string-level prediction and the character-level prediction, whether the input data is authentic input. The system transmits, to the device, a decision that is generated based on determining whether the input data is authentic.

Abstract: Techniques are disclosed relating to determining whether input data is authentic. A system detects input data, that includes text data and typing data, at a computing device. The system may generate, using a string model, a string-level prediction for the input data, where the string model is trained to increase a similarity between embeddings of authentic text data and corresponding sequences of typing data. Using a character model, the system may generate a character-level prediction for the set of input data, where the character-level model predicts an intended sequence of characters based on the text data and a sequence of typing actions included in the input data. Using machine learning, the system determines, based on the string-level prediction and the character-level prediction, whether the input data is authentic input. The system transmits, to the device, a decision that is generated based on determining whether the input data is authentic.

Abstract: There are provided systems and methods of bot detection through explainable deep learning and rule violation codebooks from generative AI. A service provider, such as an electronic transaction processor for digital transactions, may provide computing services to users for processing various requests and interactions with those users. However, malicious users may utilize bots, such as automated scripts and software applications, that attempt to conduct fraud, compromise systems and data, and the like. To provide better bot and bot activity detection, the service provider may implement an explainable deep learning system that may generate rule violations of rules indicating bot activity or presence in computing logs and interactions using a generative AI. The violations may have a corresponding explanation in codebooks to automate bot detection. When bot activity is detected, the explanation may provide a reason for the bot activity detection.

Abstract: Techniques for predicting whether a submission includes a forged image. A computer system receives a submission from a user that includes an image and image metadata, such as an identifier for the user and a User-Agent string value. An image pixel embedding is generated from the image, and a profile embedding is generated from the image metadata. The image embedding is indicative of whether the image is similar to known image forgeries. The profile embedding is generated from a user activity embedding indicative of User-Agent values associated with the user identifier. The profile embedding is generated using a machine learning model that uses stored parameters to associate user activity, device information, and forgery groups. The profile embedding thus indicates whether the user is associated with known image forgeries. The image pixel embedding and profile embedding are then used by a neural network to output a forgery prediction.

Abstract: Methods and systems are presented for bot detection. A movement of a pointing device is tracked via a graphical user interface (GUI) of an application executable at a user device. Movement data associated with different locations of the pointing device within the GUI is obtained. The movement data is mapped to functional areas corresponding to a range of the different locations of the pointing device within the GUI over consecutive time intervals. At least one vector representing a sequence of movements for at least one trajectory of the pointing device through one or more of the functional areas and a duration the pointing device stays within each functional area is generated. At least one trained machine learning model is used to determine whether the sequence of movements of the pointing device was produced through human interaction with the pointing device by an actual user of the user device.

Abstract: There are provided systems and methods of bot detection through explainable deep learning and rule violation codebooks from generative AI. A service provider, such as an electronic transaction processor for digital transactions, may provide computing services to users for processing various requests and interactions with those users. However, malicious users may utilize bots, such as automated scripts and software applications, that attempt to conduct fraud, compromise systems and data, and the like. To provide better bot and bot activity detection, the service provider may implement an explainable deep learning system that may generate rule violations of rules indicating bot activity or presence in computing logs and interactions using a generative AI. The violations may have a corresponding explanation in codebooks to automate bot detection. When bot activity is detected, the explanation may provide a reason for the bot activity detection.

Abstract: There are provided systems and methods of bot detection through explainable deep learning and rule violation codebooks from generative AI. A service provider, such as an electronic transaction processor for digital transactions, may provide computing services to users for processing various requests and interactions with those users. However, malicious users may utilize bots, such as automated scripts and software applications, that attempt to conduct fraud, compromise systems and data, and the like. To provide better bot and bot activity detection, the service provider may implement an explainable deep learning system that may generate rule violations of rules indicating bot activity or presence in computing logs and interactions using a generative AI. The violations may have a corresponding explanation in codebooks to automate bot detection. When bot activity is detected, the explanation may provide a reason for the bot activity detection.

Abstract: Methods and systems are presented for providing a framework for analyzing and classifying audio data using a split-and-merge approach. Audio data is split into multiple audio tracks that correspond to different characteristics. Each audio track is segmented, and features are extracted from each segment of the audio track. Features extracted from audio segments of each audio track is analyzed. One or more correlations between the different audio tracks are determined based on comparing features extracted from audio segments of a first audio track against features extracted from audio segments of a second audio track. The audio data is classified based on the one or more correlations.

Abstract: Techniques are disclosed that relate to predicting whether a computer-based interaction is being performed by a computer bot. A computer system may receive information describing exhibited user-presence indicators of different types that are associated with the computer-based interaction, including user-presence indicators indicative of whether the computer-based interaction is being performed by a computer bot. The computer system performs a first embedding operation to create a unified embedding that unifies the exhibited user-presence indicators into a single embedding that is representative of an aggregation of the exhibited user-presence indicators. The computer system performs a second embedding operation to create a difference embedding that is representative of a set of differences between expected user-presence indicators for the computer-based interaction and the exhibited user-presence indicators.

Abstract: Methods and systems are presented for bot detection. A movement of a pointing device is tracked via a graphical user interface (GUI) of an application executable at a user device. Movement data associated with different locations of the pointing device within the GUI is obtained. The movement data is mapped to functional areas corresponding to a range of the different locations of the pointing device within the GUI over consecutive time intervals. At least one vector representing a sequence of movements for at least one trajectory of the pointing device through one or more of the functional areas and a duration the pointing device stays within each functional area is generated. At least one trained machine learning model is used to determine whether the sequence of movements of the pointing device was produced through human interaction with the pointing device by an actual user of the user device.

Abstract: Techniques for predicting whether a submission includes a forged image. A computer system receives a submission from a user that includes an image and image metadata, such as an identifier for the user and a User-Agent string value. An image pixel embedding is generated from the image, and a profile embedding is generated from the image metadata. The image embedding is indicative of whether the image is similar to known image forgeries. The profile embedding is generated from a user activity embedding indicative of User-Agent values associated with the user identifier. The profile embedding is generated using a machine learning model that uses stored parameters to associate user activity, device information, and forgery groups. The profile embedding thus indicates whether the user is associated with known image forgeries. The image pixel embedding and profile embedding are then used by a neural network to output a forgery prediction.

Abstract: Methods and systems are presented for bot detection. A movement of a pointing device is tracked via a graphical user interface (GUI) of an application executable at a user device. Movement data associated with different locations of the pointing device within the GUI is obtained. The movement data is mapped to functional areas corresponding to a range of the different locations of the pointing device within the GUI over consecutive time intervals. At least one vector representing a sequence of movements for at least one trajectory of the pointing device through one or more of the functional areas and a duration the pointing device stays within each functional area is generated. At least one trained machine learning model is used to determine whether the sequence of movements of the pointing device was produced through human interaction with the pointing device by an actual user of the user device.