A Deep Learning Neural Network Enables Toys to Engage in Intelligent Conversation

Share Article

Hutoma just launched a kickstarter campaign for a Smart Teddy Bear, a first toy powered by a deep learning neural network that can be programmed by parents using a simple text file containing sample conversations. The technology behind the teddy bear enables kids to have a more engaging playtime while giving parents full control over what the bear can say.

Hutoma Raspberry 2 embeddable neural network

A portable deep learning network can turn an everyday object into a virtual assistant

The Smart Teddy Bear is powered by a deep learning neural network and can have a meaningful conversation with kids. Parents program what the toy says using a simple text file.

Hutoma LTD, a new London based start up has recently started a kickstarter campaign to produce a smart teddy bear powered by a deep learning neural network module that can learn to understand the human language and engage in a meaningful conversation.

Many kids are spending more and more time playing video games using dad or mum phone or tablet. While someone might argue that some video games can help stimulate the brain in a good way, it is undeniable that the proliferation of easily accessible apps and mobile phones is making harder for parents to control how much time kids spend on these devices and what they are playing with.

The My Smart Teddy Bear project was born with the idea of creating a more engaging toy that can hopefully reduce the time kids are spending with video games or television. Moreover, because every child is different, hutoma give parents total control over what kind of conversations the toy can have with their kids and what to say or not say.

The toy brain (i.e. the deep learning neural network) formulate a response based on an analysis of sample conversations stored in a text file. The kickstarter teddy bear will ship with a predefined neural network but parents can edit the default toy language or change it completely. All they need to do is create a text file with as many sample questions and answers and submit it via the phone application or web interface. The hutoma machine learning algorithm will analyses it and create a new "brain" based on the parent content.

The toy backend is also exposed to those developers interested in creating a mobile virtual assistant or digital employee. The deep learning network can be, in fact, accessed independently from the actual toy.

Share article on social media or email:

View article via:

Pdf Print

Contact Author

Maurizio Cibelli
+39 3395772596
Email >

maurizio cibelli
Email >
Visit website