Giving Machines the Ability to Think on Their Own

Just some random idea when I took the shower early this morning. Maybe someone already did this. Maybe not.

Lots of us want machines to learn by themselves. This has been studied for years, but there's not a breakthrough yet. Could there be a fundamental problem that prevents us doing so?

When deep learning came out, people went crazy about it. This is possible the future of AI, people think. However, if you look closely, the underlying structure (the number of layers, their types, etc.) all relies on our decisions, the human's decisions. This is not AI.

What if we give the machine the flexibility to also change those structures. We provide the building blocks, and they learn on their own about where to use what.

I thought of Google's AutoML. What it does basically is to automatically try many combinations of models using its powerful backend.

This is dumb but cool. To train neural nets to train neural nets. However, the trained neural network would still be a fixed neural network, meaning it does not evolve.

The most dummy solution is to do similar things like AutoML does, with a reinforcement learning like closed-loop structure. So you want the neural network (that is trained to design other neural networks) to be able to refresh its own memory.

This trained network will be a building block for one particular tasks: designing another neural network for a particular action, like object detection, language translation, etc.

This structure is clearly a layered structure. While it makes sense, most of our thinking system is not really layered. It includes many possible cooperations in different areas of the brain. So if we somehow connect these trained neural networks together, that form a larger mixture rather than layers ones, then maybe the machine could have much more flexibility that enables it to "evolve" - to think on its own.

Hmm..

Maybe.

Incomplete data causes complicated learning model?

Just an idea. Today, machine learning or deep learning largely rely on the data, the volume of data. If we have millions of images of a single object, we can train the model and eventually approximate a function that maps the image into the object. The resulting model could be complicated, requiring multiple layers of neurons and requiring days to months to train the model.

What if the complexity of modeling and training is caused by incomplete data. Here I do not mean that millions of images of a single object are not enough, are incomplete. I meant, if it is true that image of a single object is incomplete. For example, when human see a dog, running on the ground, we may obtain additional information about the dog to recognize that it is a dog, like from sound, e.g., dog barking.

Control Smartphones and Wearables Under Water

Smartphones and wearables are now waterproof. This is great. It means we can shower without taking off the smartwatch; we can swim but won't miss messages/notifications; and we can surf and still pick up phone calls even if the device dropped in water.

All these sound great, but touch screens are water-unfriendly. Whenever there is some water on the screen, the touch screen just went nuts. Either it is unresponsive or it randomly clicks like there is a ghost. Water affects the screen capacity change caused by human skin and destroy the functionalities of the touch screen.

Let's look at some alternatives. Continue reading Control Smartphones and Wearables Under Water

Turn Stale SmartPhone Into SecCamera?

I was having this idea for a long long time. Yesterday my friends and I were talking about it again. Well, apparently others have done so, if you Google it. Below is a very short list of things I found online (there are tons of apps and companies doing so).

  • https://www.camio.com/
  • http://www.makeuseof.com/tag/use-old-smartphone-tablet-security-camera/
  • https://play.google.com/store/apps/details?id=com.mantishrimp.salienteye

But it occurs to me to understand why this method isn't so popular. Many companies are making expensive security cameras, and people still buy them. Maybe because we trust things that are dedicated to what they are made for? Then I see that Yi (belonged to XiaoMi) released low-cost security camera (around $30), and a lot of people like it, including me. I realized that it isn't we don't trust low-cost solutions to leverage old smartphones for surveillance. We like them and buy them because we simply don't care, because most of us simply like new stuff.

No matter how easy the setup is for the whole "using old smartphone as monitoring system," old stuff is old. We no longer want to touch the old phones. Maybe they are painfully slow. Maybe they have been sitting in the dust for so long. Maybe some functionality in the phone do not work any more. Maybe they carry so many memories and emotionally we do not want to revisit. And we move on, we get new phones, why do we turn back?

From "We Can" To "We Should"

Just finished my cruise trip to Mexico with my gf and it was a fantastic experience, though I had to read around 100 papers and organize them for finding my potential research directions and wrote the journal.. I managed to do so before the start of new year and enjoyed the trip at the same time.. phew..

Anyway. After reading the papers about sensing and wireless at MobiCom, MobiSys, SenSys, NSDI, HotMobile from 2012 to 2015, I find the interesting phenomenon. We always claim that we can do this by using that. Taking "localization" as an example, it's been studied for years and people use all kinds of technologies (e.g., FM, Wi-Fi, RFID, sound, geomagnetic, visible light, 60GHz, etc.) to accomplish meter-level or cm-level or even mm-level accuracy. Of course, they are done by assuming various kinds of scenarios. And we show that we can do it. Most introductions would look like:  Continue reading From "We Can" To "We Should"

The Uniform and Separation of Devices

Not so long ago (maybe just several years ago), people had succeeded to unify multiple devices into one. For example, cellphone is now multi-functional and has replaced many utilities we saw in the early years. People tended to squeeze everything into a small device, and that led to the what smartphone looks like today. We don't need map anymore since there's GPS and Google Maps. Post-it turns unpopular since there are multimedia notes taking apps. There are countless things that are replaced: cameras, music stations, radios, CDs, books, flashlights, compass, and so on so forth. People literally thought a smartphone could replace everything we have in life. One recent example is ASUS's Padfone. Continue reading The Uniform and Separation of Devices