Machine Learning Technology
Our artificial deep neural network utilzes both supervised and unsupervised machine learning algorithms to train our AI agents. We leverage AWS Cloud Infrastructure and architecture to derive data into the decision by AI.
“I thought AlphaGo is merely a computer good at calculating move probabilities to win the match. However, when I saw the move 37, I was proven to be wrong. AlphaGo is indeed capable of making creative decisions. It was artistic move that ensembles the beauty of Go.”
- Lee SeDol, Go World Champion, 9 Dan
Machine Learning Technology Overview
Model Tuning
We manage hundreds of models trained on big data. Learn how we overcome the machine learning challenges of tuning our models for the optimal performance.
Recurrent Neural Network
We specialize in developing AI agents on time series big data. We structure out data nodes sequentially to form a directed graph along a temporal time series sequence for RNN.
Reinforcement Learning
Our AI agents train using reinforcement learning machine learning algorithms to take actions in an environment in order to maximize the cumulative reward.
Model Tuning
Orthogonal Optimization
One of challenges in machine learning is to find optimal set of hyperparameters to tune your model while preventing your models from overfitting.
Rather than employing naive regularization technique such as early stopping which stops the training entirely if no improvement over specified steps, we attempt to apply orthogonal model tuning optimization.
In context of model tuning, the orthogonal optimization works as follows:
fit training set well on the cost function,
fit the evaluation set well on the cost function
fit the testing set well on the cost function
perform well on real world data.
Early stopping isn’t orthogonal because it both fits the training set less well and improves the dev set performance at the same time.
Automated Hyperparameter Tuning
Due to large quantity of models we manage, it doesn’t scale to manually fine tune our models one by one. We utilize the power of AWS Cloud Infrastructure with AWS Sagemaker to automate hyperparameter tuning.
By defining a range of hyperparameters to tune within AWS Sagemaker enables us to find optimal set of hyperparameters in parallel by spawning multiple AWS EC2 instances.
Recurrent Neural Network
Long Short Term Memory (LSTM)
We utilize Recurrent Neural Network to interpret sequential time series data. By utilizing forget gates within Long Short Term Term Memory implementation, we prevent back propagated errors from vanishing or exploding.
This implementation of RNN enables us to learn tasks that can potentially require memory of thousands or maybe millions steps earlier.
Reinforcement Learning
Markov Decision Process
We engineer our reinforcement learning environment within Markov Decision Process where we define the environment, states, and actions an AI agent can take. The goal of reinforcement learning is to maximize cumulative reward as much as possible.