Natural Language Processing. Machine Learning. Deep Learning. Transformers. LSTMs. All these seem jargon to you or never got around the thousands of lines of code and ‘tricks’ to use them ? Then this post is for you.
The field of Machine Learning and Natural Language Understanding has seen rapid development in the past couple of years, thanks in no part to Deep Learning. Current AI has shown amazing leaps in performance on tasks with limited data with the help of Transfer Learning. For those who don’t know what that is… It is basically a magical tool that allows anyone to take existing AI models and train them for their own data, however small the dataset maybe. Sounds good, right ?
The harsh truth is that getting these models to work requires a substantial knowledge of coding, machine learning and deep learning. And even if you have the prerequisite knowledge, it can still be a very daunting task. Extremely daunting.
Introducing aasaan.ai – A no code platform that allows anyone, with no experience in machine learning and coding, to build, train and deploy data classifiers.
So How does Aasaan work ?
We will be using the Yelp Sentiment Classification data. However, we will not be using the labels that come with the dataset. The idea being that in real life, you will not be having labelled dataset and you may want to have custom labels. So let us begin!
Upload your dataset to aasaan.
The first step is, obviously, uploading the dataset to aasaan. At the moment, the platform only supports csv files but we plan to add many more formats in the future.
Once you upload the csv, aasaan shows you an overview of the dataset.
Now , select the column with the text that you wish to classify.
Add Labels for your dataset.
As I mentioned before, in a real world scenario, you will most probably not have access to labelled data. Or maybe, you want to have custom labels. (We do support for scenarios where your data is labelled)
At this stage, aasaan asks for all the labels/categories/classes that you wish to classify your data into. For instance, in this example, we will be using the standard sentiment labels – Positive and Negative.
For the Yelp Sentiment Classification dataset, we will be adding the Positive and Negative labels.
Labelling the dataset.
To train the model, you need to label the data. Fret not! Just label 50 examples to start with.
Once, you have labelled 50 examples using the added labels, we can move forward to the exciting bit!
After a few minutes of labelling . . . .
Train the model.
Now relax. Wait for a few minutes while aasaan trains for you.
Evaluate the model.
For the yelp dataset, the text from the test set is already preloaded.
Just press predict or on the right arrow for the next text.
For using the API or CSV prediction tab, you need to join our waiting list. It is free and we would love to have your feedback.
Not too shabby for something made in a few minutes. Such is the power of deep learning and transfer learning now you can leverage it too. Without breaking a sweat.
We have successfully built, trained and deployed a data classifier without –
- Writing a single line of code
- Configuring any gpu
- Looking at daunting documentations
- Reading and understanding machine learning papers.