Artificial Intelligence, Machine Learning and Authôt
Authôt is an innovative start-up which uses specific technologies such as Artificial Intelligence – or AI – and machine learning. It enables to create the Authôt’s automatic speech recognition technology and service. Today, we enter at the heart of the process, revealing you more about how it works and what these kinds of new technologies imply for the future.
To start with, let’s define Artificial Intelligence, Machine Learning and Deep Learning. There are differences between AI and Machine Learning (ML), and it is significant. So, what is Machine Learning? According to Tom Mitchell, it is “the study of computer algorithms that allow computer programs to automatically improve through experience”. Machine Learning is a branch of Artificial Intelligence, based on statistical learning. For instance, if you provide a Machine Learning model with many of songs you enjoy, it will be able to generate a recommendation system and suggest you music with a high percentage of probability rate that you’ll find at your taste. The system recommends the music that other users who listen to similar music liked, and that you didn’t listen to. It is the technology businesses like Netflix, YouTube or Spotify use. Now, what is artificial intelligence (AI)?
Artificial Intelligence is wide in scope.
According to Andrew Moore:
Artificial intelligence is the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.
In addition, Zachary Lipton clarifies the term AI:
is aspirational, a moving target based on those capabilities that humans possess but which machines do not.
Artificial Intelligence is constantly growing. In fact, it includes a considerable amount of different technology advances and Machine Learning is only one of them. This way, it is really important to point out that there is AI in Machine Learning, but the contrary isn’t necessarily true. Indeed, there isn’t Machine Learning within all Artificial Intelligences. The borders can be tricky.
Nowadays, tech companies use AI and ML interchangeably. Indeed, what they want is that machines can learn from experiences. In extension, Deep Learning – a type of AI that allows the machine to learn by itself in contrary to programming (that’s the learning by neutral networks requiring a lot of data to be efficient)- enabled the possibility to perform specific tasks that were impossible to do with a classic rule-based programming. Tasks within the fields of speech and face recognition, image classification or even natural language processing.
Finally, we can retain, three major great technologies that come from AI:
· Automatic generation of text: Production of text from computer data.
· Automatic Speech Recognition: Transcribing and transforming human speech into a useful format for computer applications.
· Machine Learning Platforms
In this second part, let’s have a look at the democratization of AI with voice commands or autonomous cars.
As we know it today, Artificial Intelligence is symbolized by gadgets such as Google Home, Siri (Apple) and Alexa (Amazon). Indeed, artificial intelligence technologies are becoming more and more part of our daily life. To start with, let’s talk about voice command and recognition.
For instance, SIRI. It’s a voice control computer application that includes verbal instructions given by users and responds to their requests. Developed by the American company Apple and qualified as intelligent personal assistant; it is accessible on smartphones.
Following, AI interferes also in our cars. A 100% autonomous vehicle represents without a doubt the revolution of tomorrow that completely disrupts the automotive industry.
Today, failing to possess – officially – the driving license, the artificial intelligence seconds the driver in his task. For this, it relies on a series of ultrasonic sensors, radar, cameras…If artificial intelligence can support the driver, it is also able to identify the state of well-being of the latter and its passengers.
There is also the facial recognition that identifies the driver and activate the preferences and settings. The possibilities are endless.
Authôt uses machine learning for its system of automatic transcription. Indeed, the system recognizes and learn words every day, thanks to the human proofreading requested on 60% of the files uploaded.
Authôt is the specialist of Automatic Transcription of audio/video files into text. Our innovative solution works through Automatic Speech Recognition technology.
In fact, the history of artificial intelligence is closely linked to that of translation, perhaps because knowing how to give meaning is what is most human and mysterious for the machine.
The fact is that when a user tries to translate a sentence via a website, the result is sometimes puzzling.
The automatic translation field has made spectacular progress in recent years, thanks to the use of neural network on which the machine learning and deep learning systems are based. They are the most advanced artificial intelligence technologies available today. Facebook has invested in the field, just like Microsoft, Fujitsu etc.
Neural network technology was first introduced by Google Translate via its Google Brain lab and thanks to the work of Jeff Dean, Andrew Ng…emblematic figures of artificial intelligence.
“For ten years, we used a method called ‘Phrase Based Machine Translation’ (PBMT): the algorithm cut the sentence into small pieces and we translated each little bit by adopting a statistical approach,” says Julie Cattiau, engineer and product manager at Google Translate.
It’s a bit of what a human does when he sticks too much to the text: the naturalness of the sentence is not preserved, and it can give rise to misinterpretations. With a network of neurons, on the contrary, the algorithm considers the entire sentence.
We hope that you have learned more about the possibilities of artificial intelligence and machine learning and how Authôt is directly part of technology’s future.
Authôt. You speak. We write.