Abstract: Methods, systems and apparatus, including computer programs encoded on computer storage media, for training an action selection neural network. One of the methods includes receiving an observation characterizing a current state of the environment; determining a target network output for the observation by performing a look ahead search of possible future states of the environment starting from the current state until the environment reaches a possible future state that satisfies one or more termination criteria, wherein the look ahead search is guided by the neural network in accordance with current values of the network parameters; selecting an action to be performed by the agent in response to the observation using the target network output generated by performing the look ahead search; and storing, in an exploration history data store, the target network output in association with the observation for use in updating the current values of the network parameters.

Abstract: Systems and methods to improve the robustness of a network that has been trained to convergence, particularly with respect to small or imperceptible changes to the input data. Various techniques, which can be utilized either individually or in various combinations, can include adding biases to the input nodes of the network, increasing the minibatch size of the training data, adding special nodes to the network that have activations that do not necessarily change with each data example of the training data, splitting the training data based upon the gradient direction, and making other intentionally adversarial changes to the input of the neural network. In more robust networks, a correct classification is less likely to be disturbed by random or even intentionally adversarial changes in the input values.

Abstract: An automated predictive analytics system disclosed herein provides a novel technique for industry classification. Leveraging specific API to construct a database of companies labeled with the industries to which they belong, the automated predictive analytics system trains a deep neural network to predict the industries of novel companies. The automated predictive analytics system examines the capacity of the model to predict six-digit NAICS codes, as well as the ability of the model architecture to adapt to other industry segmentation schemas. Additionally, the automated predictive analytics system investigates the ability of the model to generalize despite the presence of noise in the labels in the training set. Finally, the automated predictive analytics system explores the possibility of increasing predictive precision by thresholding based on the confidence scores that the model outputs along with its predictions.

Abstract: A system and method for creating and providing crime intelligence based on crowdsourced information stored on a blockchain, where the crowdsourced information is analyzed and evaluated preferably according to an artificial intelligence (AI) model and users are rewarded for providing timely, valuable, and accurate crime tips. The crowdsourced information may be obtained in any suitable manner, including but not limited to written text, such as a document, or audio information.

Abstract: An evolutionary AutoML framework called LEAF optimizes hyperparameters, network architectures and the size of the network. LEAF makes use of both evolutionary algorithms (EAs) and distributed computing frameworks. A multiobjective evolutionary algorithm is used to maximize the performance and minimize the complexity of the evolved networks simultaneously by calculating the Pareto front given a group of individuals that have been evaluated for multiple obj ectives.

Abstract: An artificial neural network system for managing a task to be performed by heterogeneous resources executing an artificial neural network, the artificial neural network system including a model analyzer that receives an artificial neural network model and outputs sub-graph information generated based on the artificial neural network model including at least one of sub-graph, a detector that outputs awareness information about the heterogeneous resources, and a task manager that outputs a first request signal for performing a task with respect to each layer of first resource of the heterogeneous resources based on the sub-graph information and the awareness information, and a second request signal for performing an task with respect to each depth of a second resource of the heterogeneous resources.

Abstract: An aircraft includes a propulsion system, a sensor system, a control system, and a processing system including a memory and a processor. The memory is configured to store computer-executable instructions. The processor is configured to access the memory and to execute the computer-executable instructions to perform the following steps: obtaining a set of first weights of a processing unit of a neural network; ternarizing each weight included in the set of first weights to obtain a set of second weights; generating an output of the processing unit based on the set of second weights and a set of inputs of the processing unit; and training weights included in the set of first weights of the processing unit of the neural network based on an error cost function including an error term and a structurally sparse term.

Abstract: A pipeline system for time-series data forecasting using a distributed computing environment is disclosed herein. In one example, a pipeline for forecasting time series is generated. The pipeline represents a sequence of operations for processing the time series to produce modeling results such as forecasts of the time series. The pipeline includes a segmentation operation for categorizing the time series into multiple demand classes based on demand characteristics of the time series. The pipeline also includes multiple sub-pipelines corresponding to the multiple demand classes. Each of the sub-pipelines applies a model strategy to the time series in the corresponding demand class. The model strategy is selected from multiple candidate model strategies based on predetermined relationships between the demand classes and the candidate model strategies. The pipeline is executed to determine the modeling results for the time series.

Abstract: Systems and methods for generating intents for a response is provided. The tokens of the response is encoded into a dense vector space as a plurality of vectors. Name entities are extracted, and individual sentences and paragraphs are both classified in response to the vectors. In addition to the tokens being represented in the vector space, the sentences and paragraphs may be represented in the vector space. The entities and intents are then used to determine an action for the system according to a policy that is optimized for. Annotations may be requested when the classifications are below thresholds, and these annotations may be employed in the action determination process. Annotation includes receiving an annotation work in an annotation queue, prioritizing the annotations, and sending the highest priority annotations to the annotator in order. This is used to update the production annotation database.

Abstract: This application relates to a machine learning model training method and apparatus, and an expression image classification method and apparatus. The machine learning model training method includes: obtaining a machine learning model that includes a model parameter and that is obtained through training according to a general-purpose image training set; determining a sample of a special-purpose image and a corresponding classification label; inputting the sample of the special-purpose image to the machine learning model, to obtain an intermediate classification result; and adjusting the model parameter of the machine learning model according to a difference between the intermediate classification result and the classification label, continuing training, and ending the training in a case that a training stop condition is met. The solutions provided in this application improve the training efficiency of the machine learning model.

Abstract: A system and a method provide compression and decompression of weights of a layer of a neural network. For compression, the values of the weights are pruned and the weights of a layer are configured as a tensor having a tensor size of H×W×C in which H represents a height of the tensor, W represents a width of the tensor, and C represents a number of channels of the tensor. The tensor is formatted into at least one block of values. Each block is encoded independently from other blocks of the tensor using at least one lossless compression mode. For decoding, each block is decoded independently from other blocks using at least one decompression mode corresponding to the at least one compression mode used to compress the block; and deformatted into a tensor having the size of H×W×C.

Abstract: Method and apparatus for compressing and decompressing a deep learning model. The apparatus for compressing extracts a threshold from a weight matrix for each layer of a pre-trained deep learning model, generate a binary mask for the weight matrix based on the threshold for each layer, apply the binary mask to the weight matrix for each layer of the pre-trained deep learning model, and perform a matrix sparsity process to generate a compression model.

Abstract: Techniques that facilitate model support in deep learning are provided. In one example, a system includes a graphics processing unit and a central processing unit memory. The graphics processing unit processes data to train a deep neural network. The central processing unit memory stores a portion of the data to train the deep neural network. The graphics processing unit provides, during a forward pass process of the deep neural network that traverses through a set of layers for the deep neural network from a first layer of the set of layers to a last layer of the set of layers that provides a set of outputs for the deep neural network, input data for a layer from the set of layers for the deep neural network to the central processing unit memory.

Abstract: Certain aspects of the present disclosure provide techniques for performing finite rank deep kernel learning. In one example, a method for performing finite rank deep kernel learning includes receiving a training dataset; forming a set of embeddings by subjecting the training data set to a deep neural network; forming, from the set of embeddings, a plurality of dot kernels; combining the plurality of dot kernels to form a composite kernel for a Gaussian process; receiving live data from an application; and predicting a plurality of values and a plurality of uncertainties associated with the plurality of values simultaneously using the composite kernel.

Abstract: An apparatus includes a processor to: provide a set of feature routines to a set of processor cores to detect features of a data set distributed thereamong; generate metadata indicative of the detected features; generate context data indicative of contextual aspects of the data set; provide the metadata and context data to each processor core, and distribute a set of suggestion models thereamong to enable derivation of a suggested subset of data preparation operations to be suggested to be performed on the data set; transmit indications of the suggested subset to a viewing device, and receive therefrom indications of a selected subset of data preparation operations selected to be performed; compare the selected and suggested subsets; and in response to differences therebetween, re-train at least one suggestion model of the set of suggestion models based at least on the combination of the metadata, context data and selected subset.

Abstract: A growing need for inferencing to be run on fog devices exists, in order to reduce the upstream network traffic. However, being computationally constrained in nature, executing complex deep inferencing models on such devices has been proved difficult. A system and method for partitioning of deep convolution neural network for execution of computationally constraint devices at a network edge has been provided. The system is configured to use depth wise input partitioning of convolutional operations in deep convolutional neural network (DCNN). The convolution operation is performed based on an input filter depth and number of filters for determining the appropriate parameters for partitioning based on an inference speedup method. The system uses a master-slave network for partitioning the input. The system is configured to address these problems by depth wise partitioning of input which ensures speedup inference of convolution operations by reducing pixel overlaps.

Abstract: An analog signal processing circuit comprising: a first promoter operably linked to a nucleic acid sequence encoding a first output molecule, wherein said promoter is responsive to a cooperative input signal comprising at least two cooperative inputs, and wherein expression of said at least two cooperative inputs is tunable.

Abstract: Feature selection methods and processes that facilitate reduction of model components available for iterative modeling. It has been discovered that methods of eliminating model components that do not meaningfully contribute to a solution can be preliminarily discovered and discarded, thereby dramatically decreasing computational requirements in iterative programming techniques. This development unlocks the ability of iterative modeling to be used to solve complex problems that, in the past, would have required computation time on orders of magnitude too great to be useful.

Abstract: A method includes generating, by one or more processors, a first graphical interface. The first graphical interface includes a card-based view with each card in the card-based view corresponding to a field of analysis from a plurality of fields of analysis. The method also includes transmitting, to a client device, the representation of the first graphical interface; receiving, from the client device, a selection of a particular card of the card-based view; and, based on the received selection, generating a representation of a second graphical interface that includes a detailed view of output data associated with a field of analysis that corresponds to the particular card. The method further includes transmitting, to the client device, the representation of the second graphical interface.

Abstract: A computer-implemented infrastructure providing a consistent graphical user interface that supports user-controlled organizing, storing, accessing and sharing of heterogeneous personal information of a specific user uses computer processes executed by a server system.

Abstract: A computer-implementable method for managing a cognitive graph comprising: receiving data a data source, the data comprising a query and information relating to an answer to the query; processing the query, the processing the query identifying a plurality of query related knowledge elements; processing the information relating to the answer to the query, the processing the information relating to the answer to the query identifying a plurality of answer related knowledge elements; and, storing the plurality of query related knowledge elements and the plurality of answer related knowledge elements within the cognitive graph as a collection of knowledge elements.

Abstract: A dynamic, distributed directed activity network comprising a directed activity control program specifying tasks to be executed including required individual task inputs and outputs, the required order of task execution, and permitted parallelism in task execution; a plurality of task execution agents, individual of said agents having a set of dynamically changing agent attributes and capable of executing different required tasks in said activity control; a plurality of task execution controllers, each controller associated with one or more of the task execution agents with access to dynamically changing agent attributes; a directed activity controller for communicating with said task execution controllers for directing execution of said activity control program; a communications network capable of supporting communication between said directed activity controller and task execution controllers; and wherein said directed activity controller and task execution controllers communicate via said communication net

Abstract: An apparatus, and methods of making and using, to interface with a knowledge providing user and a knowledge acquiring user(s) to provide knowledge in a domain, the apparatus in the form of a embodied computer processor, the computer processor implementing instructions on a non-transitory computer medium disposed in a database, the database in communication with the computer processor, the apparatus comprising: (1) a communication portion that provides communication between the computer processor and electronic user devices; (2) the database that contains a knowledge core; and (3) the computer processor, the computer processor performing processing including: (a) interfacing with the knowledge providing user, having knowledge in a domain area, so as to input first content related to the domain; (b) inputting second content from external sources; (c) combining the first content and the second content so as to generate combined content; (d) processing the combined content using a first neural network and generat

Abstract: A system for automated information support processing is disclosed. The system may receive a natural language information support request from a plain text input channel. The system may determine a first user intent based on the information support request. The system may compare the first user intent with a set of support rules. The system may determine a dispositioned outcome based on the set of support rules and the user intent. The system may return the dispositioned outcome.

Abstract: A method, computer program product, and computer system for applying deductive artificial intelligence (AI) attribution and auditability to data inputs, wherein the deductive AI may account for ontologies and competing system information, and wherein the deductive AI attribution and auditability may be applied to the data inputs by vendor workflow. The data inputs applied with the deductive AI attribution and auditability may be processed via a feedback loop to align a sense-understand-decide-act (SUDA) understanding with an inductive AI understanding. The inductive AI may be automated via the feedback loop based upon, at least in part, an AI expert system processing of the data inputs. One or more policy based rules may be developed for user automation authorization based upon, at least in part, the feedback loop.

Abstract: Methods and systems are provided for managing knowledge. Users of the knowledge management (KM) service or system may provide post questions, provide answers or search for answers via a user interface. For a given question, users may collectively construct and refine an answer hierarchy that may include multiple answer paths or solutions to the same question. To this end, users may add, remove and/or modify nodes of the answer hierarchy. The nodes may be of predefined types such as Step, Goto or Fork. In addition, users may provide or modify content specific to individual nodes of the answer hierarchy such as attachments, questions, references or comments. Given an answer hierarchy, users may preview alternate paths and compare different answer paths representing different solutions based on user-specific fact patterns or requirements before making their selections along a particular path to a solution.

Abstract: Systems and methods for generating a display of AI interactions in an automated conversation are provided. This display allows for simplified review of conversation flow for a user, and to also enable altering the conversation progression in an intuitive and user friendly manner. Also disclosed is managing AI transactions in the automated conversation. Systems and methods for visualizing trends in the automated conversations is also provided, as is tailoring conversations to a particular target, and provided for automatic question generation in the automated conversation. Response integration of an answer to a question in the automated conversation is also disclosed. Embodiments also disclose a Conversica Score generation and used to tune model performance within the automated conversation. Lastly, in some embodiments, systems and methods are provided for handling feedback in the automated conversation.

Abstract: Embodiments of the present systems and methods may provide techniques for measuring similarity between two datasets using classification error as a measure of the similarity between the two datasets and for improving the similarity between the two datasets. For example, in an embodiment, a computer-implemented method for determining treatment effects may comprise receiving data relating to observations of treatments outcomes of at least one treatment in a plurality of treatment groups, wherein the data for each treatment group forms a dataset, reweighting at least some of the datasets to balance biases in the data among the datasets by: determining bias between at least two datasets using a classification error; and generating balancing weights for at least one of the datasets to reduce the bias between the at least two dataset, and determining treatment effects using at least one reweighted dataset.

Abstract: Embodiments are directed to generating and training a distributed machine learning model using data received from a plurality of third parties using a distributed ledger system, such as a blockchain. As each third party submits data suitable for model training, the data submissions are recorded onto the distributed ledger. By traversing the ledger, the learning platform identifies what data has been submitted and by which parties, and trains a model using the submitted data. Each party is also able to remove their data from the learning platform, which is also reflected in the distributed ledger. The distributed ledger thus maintains a record of which parties submitted data, and which parties removed their data from the learning platform, allowing for different third parties to contribute data for model training, while retaining control over their submitted data by being able to remove their data from the learning platform.

Abstract: A rule storing unit stores a set of rules each defining facts and a relation between the facts, and weights representing importance levels of the rules. An input data storing unit stores input data. A query storing unit stores a query. An importance level calculating unit calculates an importance level of each fact type of the facts defined by the rules. A fact data generating unit generates, for each fact type, fact data in which true/false is observed as a fact by a number corresponding to the importance level of the fact type, from the input data. A fact data storing unit stores the fact data. An inference performing unit performs probabilistic inference of a result of the query by using the fact data, the rules, and the weights. An output unit outputs a result of the inference.

Abstract: A method for determining a travel destination from user generated content is proposed. The method includes determining a text string in the user generated content that indicates a place or an address. Further, the method includes determining a plurality of potential travel destinations based on the text string. The method additionally includes determining similarities between the plurality of potential travel destinations and a plurality of reference positions assigned to the user, and ranking the plurality of potential travel destinations based on the similarities.

Abstract: A virtual assistant negotiation method is provided, which includes the following steps: transmitting an event information by a first electronic device corresponding to an initiator; receiving a plurality of candidate projects generated by second electronic devices corresponding to a plurality of participants according to the event information by the first electronic devices; selecting a portion of the candidate projects to serve as recommended projects by the first electronic device of the host; and making a decision according to the opinions, for the recommended projects, corresponding to the main participant among the participants by the first electronic device.

Abstract: A method, computer program product, and computer system for applying deductive artificial intelligence (AI) attribution and auditability to data inputs, wherein the deductive AI may account for ontologies and competing system information, and wherein the deductive AI attribution and auditability may be applied to the data inputs by user workflow. The data inputs applied with the deductive AI attribution and auditability may be processed via a feedback loop to align a sense-understand-decide-act (SUDA) understanding with an inductive AI understanding. The inductive AI may be automated via the feedback loop based upon, at least in part, an AI expert system processing of the data inputs. One or more policy based rules may be developed for user automation authorization based upon, at least in part, the feedback loop.

Abstract: A rule operation request associated with a service is received by a terminal. A rule corresponding to the rule operation request is obtained from a rule set associated with the service, where the rule set has been previously obtained from the server based on at least one of a device identifier or a user identifier corresponding to the terminal. A rule tree is generated based on the rule. An operation result of the rule is determined based on the rule tree and service data related to the rule.

Abstract: Embodiments for implementing intelligent recommendations of convenient event opportunities by a processor. A group of entities may be identified for one or more event opportunities or the one or more event opportunities may be identified for the group of entities according to one or more entity selection criteria and one or more event criteria. The one or more event opportunities and the group of entities may be matched according to a level of convenience for attending the one or more event opportunities of the group of entities. The one or more matching event opportunities may be ranked and suggested to the group of entities.

Abstract: A system for answering multiple choice questions includes at least one processor configured to create a question answering model using a training data set. The system is configured to create a balanced data from the imbalanced training data set. The balancing of the imbalanced training data set is achieved by generating synthetic instances of at least one minority category, among a plurality of categories into which the training data set is categorized.

Abstract: According to an embodiment, an information processing device includes a hardware processor configured to function as: an acquisition unit configured to acquire operation statistical information on a processing circuit; a derivation unit configured to derive a memory access characteristic of the processing circuit from the acquired operation statistical information, based on a prediction model for deriving the memory access characteristic from the operation statistical information; and a determination unit configured to determine an access method from among a first access method and a second access method based on the derived memory access characteristic, the first access method transferring data in a second memory unit to a first memory unit and accessing the data in the first memory unit, the second access method accessing data in the second memory unit, an access speed of the second memory unit from the processing circuit being slower than that of the first memory unit.

Abstract: Implementations for determining deployment need for a point of interest (POI) are disclosed. In one implementation, the deployment need for a POI is determined by: receiving geographical locations of one or more users, determining, based on the geographical locations, one or more target users covered by an area to be inspected, determining one or more POI deployment need indexes of the one or more target users, a POI deployment need index of a target user being determined based on a number of POIs that have the preset function of the POI and were deployed within a set distance from the target user, and providing a total deployment need index for the area to be inspected, the total deployment need index being determined based on the one or more POI deployment need indexes.

Abstract: Methods and systems for automatically predicting the probability of regulatory compliance approval based on data contained in a data structure. A data structure can be configured to include data collated and collected from one or more regulators (e.g., regulatory agencies) and one or more value chain participants. Such data is inclusive of data indicative of actual approval-request results of applications for regulatory approval by one or more regulators and variable weights assigned to different data elements. A prediction can be then made as to the probability that an application for regulatory approval by a value chain participant will be approved by a regulator based on the data collated and collected and contained in the data structure. This approach predicts the probability that an application for regulatory approval by a value chain participant will be approved and also predicts what is missing for a regulatory approval application.

Abstract: Systems, methods, and computer-readable medium are provided for healthcare analysis. Data corresponding to a plurality of patients is received. The data is parsed to generate normalized data for a plurality of variables, with normalized data generated for more than one variable for each patient. A causal relationship network model is generated relating the plurality of variables based on the generated normalized data using a Bayesian network algorithm. The causal relationship network model includes variables related to a plurality of medical conditions or medical drugs. In another aspect, a selection of a medical condition or drug is received. A sub-network is determined from a causal relationship network model. The sub-network includes one or more variables associated with the selected medical condition or drug. One or more predictors for the selected medical condition or drug are identified.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for radio frequency band segmentation, signal detection and labelling using machine learning. In some implementations, a sample of electromagnetic energy processed by one or more radio frequency (RF) communication receivers is received from the one or more receivers. The sample of electromagnetic energy is examined to detect one or more RF signals present in the sample. In response to detecting one or more RF signals present in the sample, the one or more RF signals are extracted from the sample, and time and frequency bounds are estimated for each of the one or more RF signals. For each of the one or more RF signals, at least one of a type of a signal present, or a likelihood of signal being present, in the sample is classified.

Abstract: Embodiments of the disclosed technology concern a quantum circuit configured to implement a real time evolution unitary of a Hamiltonian in a quantum computing device, wherein a unit time evolution unitary operator is decomposed into overlapping smaller blocks of unitary operators. In some implementations, (a) the size of the overlap is proportional to the logarithm of a number of qubits in the simulated system, (b) the size of the overlap is proportional to the logarithm of a total simulated evolution time, and/or (c) the size of the overlap is proportional to a Lieb-Robinson velocity.

Abstract: Methods, systems, and apparatus for individual qubit excitation control with a global excitation drive. In one aspect, a method includes accessing a quantum system that comprises a plurality of qubits; a plurality of qubit frequency control lines, each qubit frequency control line corresponding to an individual qubit and controlling the frequency of the qubit; a driveline; a plurality of couplers, each coupler coupling a corresponding qubit to the driveline so that a plurality of qubits are coupled to the driveline; determining one or more qubits that require a rotation operation; for each qubit requiring a rotation operation: tuning the qubit frequency to the corresponding driveline frequency of the rotation operation; performing the rotation operation using a microwave pulse on the excitation drive; and tuning the qubit away from the driveline frequency of the rotation operation.

Abstract: A method for quantizing a machine learning model during an inference phase, including determining a normalization factor using a set of floating-point values and a damped value of a damped value sequence; and assigning a quantized value for each floating-point value of the set of floating-point values based on the damped value sequence and the normalization factor.

Abstract: A time-series feature extraction apparatus has a coefficient outputter to output a coefficient to be used in calculation for classifying time series data into a plurality of segments, a segment position outputter to perform calculation for classifying the time series data into the plurality of segments based on the coefficient to output information on boundary positions of the plurality of segments, a cluster classifier to classify the plurality of segments into a certain number of plurality of clusters equal to or smaller than a certain number of the plurality of segments, a representative element outputter to output a representative element which represents a local feature of each of the plurality of clusters and is set for each of the plurality of segments, a feature degree calculator to calculate a feature degree of the representative element, and a representative element updater to update the representative element based on the feature degree.

Abstract: A learning device is configured to perform learning of a decision tree. The learning device includes a plurality of learning units and a plurality of performance calculators. The plurality of learning units are configured to perform learning of the decision tree using learning data divided into pieces to be stored in a plurality of data memories. The plurality of performance calculators are each configured to calculate an index value of index values for each of the plurality of data memories, the index value indicating recognition performance of the decision tree learned by corresponding one of the plurality of learning units.

Abstract: A learning device is configured to perform learning of a decision tree by gradient boosting. The learning device includes a plurality of learning units and a plurality of model memories. The plurality of learning units are configured to perform learning of the decision tree using learning data divided to be stored in a plurality of data memories. The plurality of model memories are each configured to store data of the decision tree learned by corresponding one of the plurality of learning units.

Abstract: Described herein are embodiments of systems and a method for affective response-based user authentication. In one embodiment, a computer receives a model that was trained with data that includes: (i) temporal windows of token instances (TWOTIs) to which the user was exposed, and (ii) affective responses of the user to the TWOTIs. The computer also receives a temporal window of token instances (TWOTI) and an affective response of the user to being exposed to the TWOTI (e.g., as measured by a sensor, such as a heart rate sensor). The computer utilizes the model to calculate a predicted affective response of the user to exposure to the TWOTI, and then calculates a similarity between the affective response and the predicted affective response. The computer sends a notification indicative of the user having an affective response that is incompatible with the model, responsive to the similarity being below a threshold.

Abstract: A computer generates labels for machine learning algorithms by retrieving, from a data storage circuit, multiple label sets that contain labels that each classify data points in a corpus of data. A graph is generated that includes a plurality of edges, each edge between two respective labels from different label sets of the multiple label sets. Weights are determined for the plurality of edges based upon a consistency between data points classified by two labels connected by the edges. An algorithm is applied that groups labels from the multiple label sets based upon the weights for the plurality of edges. Data points are identified from the corpus of data that represent conflicts within the grouped labels. An electronic message is transmitted in order to present the identified data points to entities for further classification. A new label set is generated using the further classification received from the entities.

Abstract: Automated (autonomous) and computer-assisted preparation of initial training patterns for an Artificial Intelligence (AI) based automated conversational agent system, such as an AI-based chatbot, includes a computer processor accessing a corpus of digital weighted conversation models representing text-based interlocutory conversations, wherein each digital weighted conversation model contains annotations and paths, and wherein each path in each digital weighted conversation model is associated with a weight; selecting a plurality of the conversations which meet at least one criteria and in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models; converting the weights associated with the selected conversations into initial training pattern values according to at least one Artificial Intelligence (AI) based automated conversational agent system; and exporting the training pattern values to at least one Artificial Intelligence (AI) based aut