How to manage a Machine Learning project?
Machine Learning is the new tool to achieve software 2.0 of a product by shifting it from rule-driven to data-driven. Machine Learning enables the apps and services to go from only interactive to personalized.
Despite how hot and trendy of a topic Machine Learning is, in reality, the failure rates…
India is a culture rich ancient civilization that has sustained and evolved through a journey of thousand years. A civilization which adopted the farming 5000 years back, had smarter architect city like Harappa, Mohenjodaro as early as 3200BC around. Around the 5th century, when the world community was learning to…
A concise analysis of roles and types of activation functions in neural networks. Detailed pros and cons of 5 popularly used ones in an Artificial Neural Network.
Activation function, as the name suggests, decides whether a neuron should be activated or not based on the addition of a bias with the weighted sum of inputs. Hence, it is a very significant component of Deep Learning, as they in a way determine the output of models. …
We vectorize input texts using pre-trained Albert embeddings from tensorflow-hub. It is used as a Keras layer, so can be extended to build deep learning models easily.
The advent of BERT has disrupted the traditional paradigm of NLP. The downstream model building is now from a model primed with knowledge of a language rather than from scratch. ALBERT can be called a lite BERT with a greatly reduced number of parameters that use a transformer-encoder architecture. In…
We will build a sarcasm classifier for news headlines using the FastText python module.
FastText is a library created by the Facebook Research Team for efficient learning of word representations and sentence classification. It has gained a lot of attraction in the NLP community especially as a strong baseline for word representation replacing word2vec as it takes the char n-grams into account while getting…
India is the fastest growing economy and the largest democracy of the world with the potential to become the next superpower in a coming couple of decades. A country like India where women are the binding force of the very culture and society, the empowerment of women is the empowerment…
Last month Tensorflow 2 got released making the model development and deployment easier with tight integration of high level keras, cleaning up of redundant APIs, keeping eager execution as default, removal of global and usage of function instead of sessions.
In this article, we will build a CNN classifier for…
Real world applications requires number of services, micro services with different service having different kind of dependencies. So It is very essential to setup separate development environment for separate projects and even services to make the development flexible. …
One of the biggest underrated challenges in machine learning development is the deployment of the trained models in production that too in a scalable way. One joke on it I have read is “Most common way, Machine Learning gets deployed today is powerpoint slides :)”.
Why Docker?
Docker is a containerization platform…
Last year, the major developments in “Natural Language Processing” were about Transfer Learning. Basically, Transfer Learning is the process of training a model on a large-scale dataset and then using that pre-trained model to process learning for another target task. …
Sambit Mahapatra
Putting ML to Customer Support at CSAT.AI | Natural Language Processing | Full Stack Data Scientist (sambit9238@gmail.com)
