Codeless ML with TensorFlow

Advances in AI frameworks enable developers to create and deploy deep learning models with as little effort as clicking a few buttons on the screen. Using a UI or an API based on Tensorflow Estimators, models can be built and served without writing a single line of machine learning code.

Blockdrop to Accelerate Neural Network training by IBM Research

IBM Research, with the help of the University of Texas Austin and the University of Maryland, has created a technology, called BlockDrop, that promises to speed convolutional neural network operations without any loss of fidelity. This could further excel the use of neural nets, particularly in places with limited computing capability. Increase in accuracy level have been accompanied by increasingly complex and deep network architectures. This presents a problem for domains where fast inference is essential, particularly in delay-sensitive and realtime scenarios such as autonomous driving, robotic navigation, or user-interactive applications on mobile devices. Further research results show regularization techniques for fully connected layers, is less effective for convolutional layers, as activation units in these layers are spatially correlated and information can still flow through convolutional networks despite dropout.

Intel’s AI Lab Presents Several Ground-breaking Research Papers

Researchers at Intel’s AI Lab recently presented several compelling research papers at the International Conference on Machine Learning (ICML) (June 10-15) and the Conference on Computer Vision and Pattern Recognition (CVPR) June 16-20.
• Collaborative Evolutionary Reinforcement Learning for Robotics and More
• Rethinking ‘Training’ of Neural Nets
• Introducing PartNet: The First Large-Scale Dataset for 3D Objects
• Leveraging Acoustics For Digital Imaging
• Deeply-supervised Knowledge Synergy for Advancing the Training of Deep Convolutional Neural Networks

Use the k-means clustering, Luke

In my last post I scraped some character statistics from the mobile game Star Wars: Galaxy of Heroes. In this post, I’ll be aiming to try out k-means clustering in order to see if it comes out with an intuitive result, and to learn how to integrate this kind of analysis into a tidy workflow using broom.

Visualize monthly precipitation anomalies

Normally when we visualize monthly precipitation anomalies, we simply use a bar graph indicating negative and positive values with red and blue. However, it does not explain the general context of these anomalies. For example, what was the highest or lowest anomaly in each month? In principle, we could use a boxplot to visualize the distribution of the anomalies, but in this particular case they would not fit aesthetically, so we should look for an alternative. Here I present a very useful graphic form.

Deep Learning for NLP: ANNs, RNNs and LSTMs explained!

Learn about Artificial Neural Networks, Deep Learning, Recurrent Neural Networks and LSTMs like never before and use NLP to build a Chatbot!

Dear Random Forest

Lately, I’ve been thinking a lot about you, wondering how you are feeling, and pondering about what the future has for you. You see, I’m sad that the sound of your leaves rustling is not being heard the way it used to. Since some years ago, your other fans and practitioners decided to leave the trails of your forest in exchange for a more neural and networked life. And while I’m sure they had their reasons, this makes me sad. Let me tell you something.


The process of pulling in a member of society (often called user by technology companies) into submitting their data is becoming a common practice. A fun activity that pulls you in is contrasted by the sale of user’s data that occurs without clear consent. I suggest with this article to use the shorthand phrase databaiting for this process. Who buys your data is usually not something you know when you join a service or buy a product, rather it is a process of discovery. An exciting user journey? In writing this article it is not my intention to offend, rather to make you more conscious of the way the data you contribute is being used. It would be great too if you started questioning the companies you give your data and what data is in the first place. We could perhaps talk of responsible contribution of data, yet I am unsure of how that would function in practice. Before defining the term let us run through a few short examples, for fun of course, and see what you think.

Support Vector Machines for Classification

Learn about Support Vector Machines (SVM), from intuition to implementation.

Reinforcement Learning: beyond the supervised and unsupervised ways

The idea is that we have an agent(the robot) and an environment (the labyrinth). That agent has a personal representation of the environment, which is called state. Then, it needs to interact with the environment and the way it does that is through actions. After each action, the agent will have a new state of the environment, as well as feedback on its action, in terms of reward or penalty.

Log Book – Guide to Hypothesis Testing

This is a guide to Hypothesis testing. I have tried to cover the basics of theory and practical implementation with a step by step example.

Making Grab’s everyday app super

Grab is Southeast Asia’s leading superapp, providing highly-used daily services such as ride-hailing, food delivery, payments, and more. Our goal is to give people better access to the services that matter to them, with more value and convenience, so we’ve been expanding our ecosystem to include bill payments, hotel bookings, trip planners, and videos – with more to come. We want to outserve our customers – not just by packing the Grab app with useful features and services, but by making the whole experience a unique and personalized one for each of them. To realize our super app ambitions, we work with partners who, like us, want to help drive Southeast Asia forward.

Regular Expressions Explained

I bet you all have encountered regular expressions at some points. They are very powerful tools that are universally supported in many platforms, including programming languages like Python, R, Java, SQL, Scala. As a data scientist/developer, having a solid understanding of Regex can help you perform various data munging and text mining tasks very easily. Personally, I use them for lots of random stuffs, mostly when I have to work with text data or do Natural Language Processing projects.

Beyond Graph Convolution Networks

Graph Neural Networks, though a fairly new concept, has gained immense popularity as an interesting methodology of analyzing graphs, the deep learning way. Being able to naturally fit many real-world datasets, which have an inherent graph structure, GNNs have found applications in many different domains from online advertising to road traffic prediction and drug designing. This blog post: Applications of Graph Neural Networks enumerates the variety of areas where GNNs have achieved significant results. The early GNNs would learn a target node’s representation by propagating neighbour information in an iterative manner until a stable fixed point is reached. However, GNNs hardly stayed in this nascent form, quickly adopting ideas from other successful areas of deep learning to evolve to the architecture we know today as Graph Convolution Networks.