How Much Deeper Can Deep Learning Goes?
- Choo Boon Linn
- Oct 27, 2020
- 3 min read
Updated: Oct 29, 2020
Is a wall hit by deep learning model today really an obstacle? I don't think so. But how much deeper it can go?

This is a question equivalent to how is the deepest level of Ocean looks like? Is there any living form in the deepest zone of Earth crust? How many stars are there in the universe? Do parallel universe really exists? Where we go after-life?
I Don't Know
But I have suggestions what we can try out with it.
1) Younger population is living more luxurious lives than older generation. Many times, this also bring us hard times, for being unwise with our decisions in spending. With the rich working to make more money out from the consumers without much concerning the impacts post-allured to the customers, we need a good AI advisor companion. One of which is not solely working to fish benefits from you but to help you in living to the very best you can be. What is more tempting than having a shadow companion that can advise how a human can make a decision in improving one’s economy in less efforts manner but promising results while also allowing more family time as possible?
2) Okay, we are smart enough to create technology. But we are also dummy enough to let someone abused the technology invested. How many persons out there lost their lives due to cyber bully and insane challenges? Hello friends, why do we not working to make deep learning tackles all these cyber bullying? Is deep learning just a medium bringing profits to the inventor? No, deep learning does not have a gender or bias, it can do more than that. It can save humanity. Do not overlook the need to balancing out the cons and pros. We got to act fast.
3) During an investigation, we always have to interview multiple parties. Can their stories be combines by AI to build a more complete story and spot the contradiction?
4) Reverse engineering is an honorable term. Can deep learning validates our research pipelines from the opposite way? For instance, can we reconstruct subject's perceived images via the collected fMRI or EEG signals, or even activated spatial cells?
Next, Recollecting Languages
Not every language and dialect is still intact. But there are intimate relationships between multiple languages. Can deep learning figure it all out?!
And oh, aging!
Cognitive decline is not the only issue in aging
It is not just slower in information processing and movement, or poorer memory, aging also accompanied with a deterioration in eyesight, hearing, sensing and smelling! What if some elderly are still using gas cooker and never able to detect a leakage of gas? Google Nose announced on April Fool Day of 2013 can actually be a great idea. If we can create a system that can analyze the component and concentration of air particles in real time, we can alert emergency in no time for the sake of every live matters.
Okay, but first, transparency AI
We have theories over development of brain through the biological models. And we are all excited with the outcome brought by deep learning. But how it reaches the outcome in the first place? It is a black box. Hence, everyone is talking about transparency of it now. To avoid a nightmare on how Earth may eventually get terrorized by AI per suggested by some? Ahhh, hmmmm, I won't know. But yeah, okay, this is science, we want to know how A goes up till Z.
I don't have an exclusive idea in doing this yet. But perhaps, this experiment have to work with different settings. Corresponding a model development to a brain development, unsure if we can see: (i) hardware specifications as opposed to growth factors, (ii) resistance, currents and temperature to growth environment, (iii) base logic to DNA, while every result may show a specific pathway with different background (initial state to goal) having different circuit preference? Hmmmm, well, okay. Let me brainstorm over this further and come back again sometime. Stay tune!
Comments