Abstract: A processor may receive a plurality of text samples generated by a user and identify at least one variable text element in at least one of the plurality of text samples. The processor may tokenize the at least one variable text element, thereby producing a plurality of tokenized text samples including at least one token. The processor may build a longest common substring from the plurality of tokenized text samples and add the longest common substring and the at least one token to a set of selectable user interface options specific to the user. The processor may generate a user interface comprising the set of selectable user interface options. This can include detecting a user interface context and automatically replacing the at least one token with information specific to the user interface context within the set of selectable user interface options.

Abstract: Aspects of the present disclosure provide techniques for detecting latent bias in machine learning models. Embodiments include receiving a data set comprising features of a plurality of individuals. Embodiments include receiving identifying information for each individual of the plurality of individuals. Embodiments include predicting, for each respective individual of the plurality of individuals, a probability that the respective individual belongs to a given class based on the identifying information for the given individual. Embodiments include providing, as inputs to a machine learning model, the features of the plurality of individuals from the data set. Embodiments include receiving outputs from the machine learning model in response to the inputs. Embodiments include determining whether the machine learning model is biased against the given class based on the outputs and the probability that each respective individual of the plurality of individuals belongs to the given class.

Abstract: Feature randomization for securing machine learning models includes receiving an event, and altering, responsive to receiving the event, a threshold pseudo-randomly to generate an altered threshold value. Feature randomization further includes applying the altered threshold value to a threshold-dependent feature to generate an altered threshold-dependent feature value. The altered threshold-dependent feature value determined at least in part from the event. Feature randomization further includes executing a machine learning model, on the event and the altered threshold-dependent feature value, to generate a predicted event type for the event.

Abstract: A plurality of graph snapshots for a plurality of consecutive periodic time samples maps between connected components in consecutive graph snapshots and describes at least one feature of each connected component. A recursively-built tree tracks an evolution of one of the connected components through the plurality of graph snapshots, the tree including a root node representing the connected component at a final one of the consecutive periodic time samples and a plurality of leaf nodes branching from the root node. A plurality of paths is extracted from the tree by traversing the tree from the root node to respective ones of the plurality of leaf nodes. Each path contains data describing an evolution of a respective one of the connected components through time as indicated by evolution of the at least one feature of the respective one of the connected components.

Abstract: Aspects of the present disclosure involve systems, methods, devices, and the like for auto-labeling clusters generated by machine learning models. In one embodiment, a system is introduced that can perform a series of operations for determining comprehensive labels for clusters output from machine learning methods used to classify data sets. The auto-labeling system may include generating labels determined using a computation of a frequency count, ratio, and coverage. These computations may use feature-based dictionaries which aid in the determination, storage, and analysis of the relevant features useful in labeling the clusters.

Abstract: Aspects of the present disclosure involve systems, methods, devices, and the like for auto-labeling clusters generated by machine learning models. In one embodiment, a system is introduced that can perform a series of operations for determining comprehensive labels for clusters output from machine learning methods used to classify data sets. The auto-labeling system may include generating labels determined using a computation of a frequency count, ratio, and coverage. These computations may use feature-based dictionaries which aid in the determination, storage, and analysis of the relevant features useful in labeling the clusters.

Abstract: Techniques are disclosed relating to validating cluster results. A computer system may receive a first cluster result generated at a first computer platform configured to execute a first software implementation of a clustering algorithm to generate the first cluster result. The first cluster result may include a first set of clusters, each of which groups one or more of a plurality of data values. The computer system may receive a second cluster result generated at a second computer platform configured to execute a second, different software implementation of the same clustering algorithm to generate the second cluster result. The second cluster result may include a second set of clusters. The computer system may cause cluster information to be presented to a user that indicates that a cluster of the first set of clusters groups data values that are grouped by two or more of the second set of clusters.

Abstract: Techniques are disclosed relating to validating cluster results. A computer system may receive a first cluster result generated at a first computer platform configured to execute a first software implementation of a clustering algorithm to generate the first cluster result. The first cluster result may include a first set of clusters, each of which groups one or more of a plurality of data values. The computer system may receive a second cluster result generated at a second computer platform configured to execute a second, different software implementation of the same clustering algorithm to generate the second cluster result. The second cluster result may include a second set of clusters. The computer system may cause cluster information to be presented to a user that indicates that a cluster of the first set of clusters groups data values that are grouped by two or more of the second set of clusters.

Abstract: Aspects of the present disclosure involve systems, methods, devices, and the like for auto-labeling clusters generated by machine learning models. In one embodiment, a system is introduced that can perform a series of operations for determining comprehensive labels for clusters output from machine learning methods used to classify data sets. The auto-labeling system may include generating labels determined using a computation of a frequency count, ratio, and coverage. These computations may use feature-based dictionaries which aid in the determination, storage, and analysis of the relevant features useful in labeling the clusters.