What could be the most important field of research in artificial intelligence? Natural language understanding may be the one. Advances in natural language understanding can potentially boost up the productivity and knowledge acquisition for humans.

Human and machine work together: Two important intelligent applications

In 2018, I started a journey into the world of cryptocurrency trading and software development with minimal experience. Alongside a knowledgeable friend, I learned new concepts in trading, cryptocurrency, and software development. Our adventure led us through highs and lows, culminating in valuable lessons that reshaped our approach to technology and finance.

Lessons from my first trading journey in 2018 — Can arbitrage make me rich?

Artificial neural network (deep learning may be a fancier name!) is probably the most popular machine learning model now. You can always hear the words “deep learning” from everywhere. This is a biologically inspired model and it is good to understand this model with some biology!

Understand artificial neural networks with biological analogy

Memory is the data or information retrieved when an intelligent agent needs it. The ability to retrieve and store memory is one of the features of intelligent agents. With the theory of Complementary Learning Systems (CLS), an intelligent agent implements 2 learning systems— one in the neocortex for slow learning, one in the hippocampus for fast learning. Learning in a deep artificial neural network can be considered as slow learning in the neocortex. Training a deep neural network is data-inefficient, even though a deep neural network is powerful. A Release from DeepMind reviewed CLS and proposed that there may be insight from this theory to address this problem.

What Learning Systems do Intelligent Agents Need? Reviews from DeepMind 1

We know that there are 2 learning systems. But why do we need 2 separate learning systems? In this article, we will explain the role of the hippocampus to the slow integration of new experiences through replay. This provides insight for us to tackle the problem we called ‘catastrophic forgetting’ in training an artificial neural network. The quick adaption and slow integration of new experiences mechanism suggested in CLS may also provide insights on designing the architectures of an intelligent agent.

Fisher Lok

Blog

Human and machine work together: Two important intelligent applications

Lessons from my first trading journey in 2018 — Can arbitrage make me rich?

Understand artificial neural networks with biological analogy

What Learning Systems do Intelligent Agents Need? Reviews from DeepMind 1

What Learning Systems do Intelligent Agents Need? Reviews from DeepMind 2