International Business Machines Corporation

Akira Koseki

Hideaki Komatsu

Toshiro Takase

Kiyokuni Kawachiya

Hiroki Yanagisawa

Tamiya Onodera

Takayuki Katsuki

Tetsuro Morimura

Mikio Takeuchi

Toshiya Iwamori

Takeshi Imamura

Yuji Watanabe

Motohiro Kawahito

Tadanobu Inoue

Kohji Takano

Masaki Ono

Toru Nagano

Takao Moriyama

Hiroshi Kanayama

Tatsushi Inagaki

Toshio Suganuma

Haruki Imai

Hiroaki Nakamura

Takayuki Osogami

Hiroshi Horii

Takeshi Ogasawara

Michiaki Tatsubori

Toshihiro Takahashi

Taiga Nakamura

Kazuaki Ishizaki

Hironori Takeuchi

Graeme Johnson

Kohtaroh Miyamoto

Akihiko Tozawa

Koichi Takeda

Toshinari Itoko

Akihiro Kosugi

Kohei Miyaguchi

Hiroyuki Momose

Hiroshi Horri

A computer-implemented method for updating hidden states in a recurrent neural network (RNN) to predict future data from multivariate time-series data with irregular time intervals is provided including inputting, for each of time steps at observations, observation data at a current time step in the multivariate time-series data to the RNN, for each of the time steps: subdividing a time interval between a previous time step and the current time step by a predetermined number, for each of subdivided time steps calculating a first element of the hidden state at a current subdivided time step using ODE-RNNs, and calculating a second element of the hidden state at the current subdivided time step using the last updated hidden state and a hidden state at the previous time step so that the last updated hidden state is decayed to be close to the hidden state at the previous time step.

Neural ODE for irregularly-sampled time series

A computer-implemented method is provided for predicting a medical event time. The method includes receiving an electronic health record (EHR) including a plurality of pairs of observation variables and corresponding timestamps. Each of the plurality of pairs include a respective observation variable and a respective corresponding timestamp. The method further includes converting the EHR into a K-dimensional vector representing a cumulative-stay time at a finite number of patient medical states, the patient medical states being determined by values of the observation variables. The method also includes processing, by a hardware processor, the K-dimensional vector using a medical event time prediction model to output a prediction of a medical event time. The medical event time prediction model has been configured through training to receive and process K-dimensional vectors converted from past EHRs to output predicted medical event times.

Diabetes complication prediction by health record monitoring

A computer-implemented method, a computer program product, and a computer system for using a time-window based attention long short-term memory (TW-LSTM) network to analyze sequential data with time irregularity. A computer splits elapsed time into a predetermined number of time windows. The computer calculates average values of previous cell states in respective ones of the time windows and sets the average values as aggregated cell states for the respective ones of the time windows. The computer generates attention weights for the respective ones of the time windows. The computer calculates a new previous cell state, based on the aggregated cell states and the attention weights for the respective ones of the time windows. The computer updates a current cell state, based on the new previous cell state.

Time-window based attention long short-term memory network of deep learning

A computer-implemented method is provided for machine prediction. The method includes forming, by a hardware processor, a Convolutional Dynamic Boltzmann Machine (C-DyBM) by extending a non-convolutional DyBM with a convolutional operation. The method further includes generating, by the hardware processor using the convolution operation of the C-DyBM, a prediction of a future event at time t from a past patch of time-series of observations. The method also includes performing, by the hardware processor, a physical action responsive to the prediction of the future event at time t.

Convolutional dynamic Boltzmann Machine for temporal event sequence

A computer-implemented method for learning a policy for selection of an associative topic, which can be used in a dialog system, is described. The method includes obtaining a policy base that indicates a topic transition from a source topic to a destination topic and a short-term reward for the topic transition, by analyzing data from a corpus. The short-term reward may be defined as probability of associating a positive response. The method also includes calculating an expected long-term reward for the topic transition using the short-term reward for the topic transition with taking into account a discounted reward for a subsequent topic transition. The method further includes generating a policy using the policy base and the expected long-term reward for the topic transition. The policy indicates selection of the destination topic for the source topic as an associative topic for a current topic.

Learning of policy for selection of associative topic in dialog system

A method is provided for anonymizing statistical data for a secure transfer. The method calculates statistical information for each of the statistical data. The method aggregates the statistical information to calculate a valid range for each of the statistical information. The method removes outlier data based on the valid range for each of the statistical data. The method creates pair lists from each of the statistical data and target data, the pair lists having a respective member from both the statistical data and the target data. The method replaces each respective member of the target data by a random number existing in a range of a corresponding one of a plurality of target data bins. The method swaps each pair in each pair list in a random order using the randomized number, wherein the random number used for swapping is different for different ones of the pair lists.

Preserving privacy for data analysis

A method for implementing artificial intelligence agents to perform machine learning tasks using predictive analytics to leverage ensemble policies for maximizing long-term returns includes obtaining a set of inputs including a set of ensemble policies and a meta-policy parameter, selecting an action for execution within the system environment using a meta-policy function determined based in part on the set of ensemble policies and the meta-policy function parameter, causing the artificial intelligence agent to execute the selected action within the system environment, and updating the meta-policy function parameter based on the execution of the selected action.

Implementing artificial intelligence agents to perform machine learning tasks using predictive analytics to leverage ensemble policies for maximizing long-term returns

A method for selecting an action, includes reading, into a memory, a Partially Observed Markov Decision Process (POMDP) model, the POMDP model having top-k action IDs for each belief state, the top-k action IDs maximizing expected long-term cumulative rewards in each time-step, and k being an integer of two or more, in the execution-time process of the POMDP model, detecting a situation where an action identified by the best action ID among the top-k action IDs for a current belief state is unable to be selected due to a constraint, and selecting and executing an action identified by the second best action ID among the top-k action IDs for the current belief state in response to a detection of the situation. The top-k action IDs may be top-k alpha vectors, each of the top-k alpha vectors having an associated action, or identifiers of top-k actions associated with alpha vectors.

Partially observed Markov decision process model and its use

A method is provided for rule creation that includes receiving (i) a MDP model with a set of states, a set of actions, and a set of transition probabilities, (ii) a policy that corresponds to rules for a rule engine, and (iii) a set of candidate states that can be added to the set of states. The method includes transforming the MDP model to include a reward function using an inverse reinforcement learning process on the MDP model and on the policy. The method includes finding a state from the candidate states, and generating a refined MDP model with the reward function by updating the transition probabilities related to the state. The method includes obtaining an optimal policy for the refined MDP model with the reward function, based on the reward policy, the state, and the updated probabilities. The method includes updating the rule engine based on the optimal policy.

Rule creation using MDP and inverse reinforcement learning

A computer-implemented method is presented for building a topic model to discover topics in a collection of documents generated by a plurality of users. The method includes extracting conversations from the collection of documents, dividing the extracted conversations into a plurality of segments, generating a topic distribution for each of the plurality of segments based on the extracted conversations and a first pre-defined prior probability distribution, and generating continuous value constructs for each of the topic distributions based on an external corpus and a second pre-defined prior probability distribution, wherein similarity is defined between the continuous value constructs.

Interactively building a topic model employing semantic similarity in a spoken dialog system

A computer-implemented method for constructing and searching structured data of question sentences in a question-answer database using machine learning and natural language processing includes receiving one or more question-answer articles as input from one or more question-answer databases over at least one network, extracting one or more profile keywords from a profile associated with a user, assigning a personalization score corresponding to the user to each of the one or more question-answer articles based on the one or more profile keywords, including applying a morphological and dependency analysis to a body section of the question-answer article, receiving a query for obtaining question-answer data relevant to the user, and outputting question-answer data based at least in part on the personalization score of each question-answer article in satisfaction of the query.

Constructing and searching structured data in a question-answer database using machine learning and natural language processing

A method, a system, and a computer program product provide for changing a boundary between zones. This includes setting a location of a boundary which defines a part of a zone; detecting that an object moves into the zone across the boundary; and in response to the detection, changing the location of the boundary in a direction away from the object, compared to the location of the boundary before the change.

Dynamic boundary setting

Growing community of inventors

Sagamihara, Japan

Akira Koseki