Jump to content

Machine Learning / Artificial Intelligence


F-Lexx

Recommended Posts

Is anyone here into Machile Learning or Artificial Intelligence? I'd be curious about what you're working on and what methods and tools you use (simulation software, libraries, programming languages etc.).

 

Personally, I'm doing research into modelling and simulation of gene regulation networks (GRNs). I'm basically trying to infer the topology of a graph of genes from dynamic (temporal) gene expression data. The GRN models consist of ODE systems where each equation depicts the gene expression kinetics of a particular gene, dependig on the influences from other genes. I use a natural computing algorithm called Particle Swarm Optimization (PSO) to infer the parameters of these models.

 

I created my own tool from scratch using Java, despite the fact that it's intended to be executed on HPC clusters and grids. Computation speeds are quite good, especially due to parallelization -- I'm using a multi-swarm variant of PSO, where each swarm is simulated independently and the local optima are synchronized every 100ish iterations.

Link to comment
Share on other sites

One of the modules on my course this year/next year-I'm repeating due to medical issues-is on AI, and I must admit, it's a rather new area for me. It's certainly an interesting area of research, but I don't know if we'll truly have AI for some time. It's strange. At our current level of technology-and someone can correct me on this obviously-we can only create an AI which learns within certain parameters. So for example, we can create a voice recognition system which can adapt to the different accents of a person, or 'learn' what you mean by certain words which are outside of its preset definitions. 

 

We can't however create an AI which can learn to adapt outside of the parameters which are set by its creators. So for example, we can't...I dunno, ask Siri what its opinion of the best restaurants for us would be in a given area. It can only say, 'Well I was programmed to say this'. Or am I wrong?

Link to comment
Share on other sites

  • Administrators

We can't however create an AI which can learn to adapt outside of the parameters which are set by its creators. So for example, we can't...I dunno, ask Siri what its opinion of the best restaurants for us would be in a given area. It can only say, 'Well I was programmed to say this'. Or am I wrong?

Hmm..kind of.  But you don't have to hard code every answer.  The programming behind that question would be:

 

1. Go Google best restaurants in current city.

2. Compile reviews

3. Look for highest reviews

4. Return back restaurant name

 

This way yes it's following logic, but it's aggregating real data to come to it's conclusion. 

Link to comment
Share on other sites

I know of that, but I mean the program-as far as I know-can't adapt to what you have liked in the past. Well, it can of course since...you could just code it to 'like' a restaurant if the user says,'Ohh that sounds nice, let's go there' I suppose. 

 

Honestly, I know what I mean to say, I just can't think of how to say it in words. Which is odd.

Link to comment
Share on other sites

Nathan, that seems like a very interesting project. The market seems quite auspicious now for virtual personal assistants now. Best of luck with it, I hope you'll make a couple more milestones before the project gets funded.

 

Livvy, you're probably thinking about "weak AI" vs. "hard AI". If so, then I agree that most work done today is towards developing weak AI, and that's perfectly OK, lots of practical applications are currently coming out of this. Me, I'm more interested in hard AI, also called Artificial General Intelligence (AGI). There's quite a bit of interesting work done in artificial cognition and artificial consciousness, although it wouldn't be a shock to learn that these fields are quite immature at the moment. If you're curious about this stuff, there's a very nice open-source project called OpenCog that I'm really excited about.

Link to comment
Share on other sites

It's funny you brought this topic up.  I just pledged on kickstarter for an AI project.  It's over their funding goal and still climbing.

 

https://www.kickstarter.com/projects/181239886/jaesa

 

I am a total newbie in what concerns artificial intelligence, but it's something that interests me a lot. I am going to print this article you post for detailed reading, thanks for sharing! I don't know up to what point there is a connection between AI and the stuff we see in the movies regarding the machines taking over...

Link to comment
Share on other sites

I am a total newbie in what concerns artificial intelligence, but it's something that interests me a lot. I am going to print this article you post for detailed reading, thanks for sharing! I don't know up to what point there is a connection between AI and the stuff we see in the movies regarding the machines taking over...

That depends on which movies you're referring to. If you're thinking of Her, that would be something like Nathan's project, with advanced cognitive abilities. If you've seen Transcendence, that would take things up a notch. Their version of AI has access to very powerful hardware. This gives it control over a wide array of devices which affect our lives more or less. The movie's plot touches on some interesting philosophical and technological issues, namely mind uploading and the technological singularity, which is defined on Wikipedia as such: "a hypothetical moment in time when artificial intelligence, human biological enhancement, or brain-computer interfaces will have progressed to the point of a greater-than-human intelligence, radically changing civilization, and perhaps human nature."

Link to comment
Share on other sites

Is anyone here into Machile Learning or Artificial Intelligence? I'd be curious about what you're working on and what methods and tools you use (simulation software, libraries, programming languages etc.).

 

Personally, I'm doing research into modelling and simulation of gene regulation networks (GRNs). I'm basically trying to infer the topology of a graph of genes from dynamic (temporal) gene expression data. The GRN models consist of ODE systems where each equation depicts the gene expression kinetics of a particular gene, dependig on the influences from other genes. I use a natural computing algorithm called Particle Swarm Optimization (PSO) to infer the parameters of these models.

 

I created my own tool from scratch using Java, despite the fact that it's intended to be executed on HPC clusters and grids. Computation speeds are quite good, especially due to parallelization -- I'm using a multi-swarm variant of PSO, where each swarm is simulated independently and the local optima are synchronized every 100ish iterations.

What type of rig are you running this java app on? I was under the impression that gene expression apps ran way better under something like Fortran rather than such a language as Java. If your running this on a desktop, how many graphics cards do you have in your rig and are you taking advantage of them to speed up processing?

Link to comment
Share on other sites

Livvy, you're probably thinking about "weak AI" vs. "hard AI". If so, then I agree that most work done today is towards developing weak AI, and that's perfectly OK, lots of practical applications are currently coming out of this. Me, I'm more interested in hard AI, also called Artificial General Intelligence (AGI). There's quite a bit of interesting work done in artificial cognition and artificial consciousness, although it wouldn't be a shock to learn that these fields are quite immature at the moment. If you're curious about this stuff, there's a very nice open-source project called OpenCog that I'm really excited about.

Yep, that's why I was getting confused!

 

I was thinking of hard AI, and trying to describe that we can't really just whip that up in the average computer lab, since at the moment-again as far as I know-we can only really create a weak AI which does as Nathan described;take results from elsewhere or from a preset number of definitions as it were. It can't really 'think' for itself, only has the appearance of thinking. I know some of the professors at my uni are working on solving this, and more importantly, trying to solve the biggest problem of all;how on earth do we begin thinking in the correct manner which would lead us to understanding how we could ever code what is essentially, a sentient being.

Link to comment
Share on other sites

That depends on which movies you're referring to. If you're thinking of Her, that would be something like Nathan's project, with advanced cognitive abilities. If you've seen Transcendence, that would take things up a notch. Their version of AI has access to very powerful hardware. This gives it control over a wide array of devices which affect our lives more or less. The movie's plot touches on some interesting philosophical and technological issues, namely mind uploading and the technological singularity, which is defined on Wikipedia as such: "a hypothetical moment in time when artificial intelligence, human biological enhancement, or brain-computer interfaces will have progressed to the point of a greater-than-human intelligence, radically changing civilization, and perhaps human nature."

 

Yeah, I've seen Transcendence recently and I find that hard to believe to happen in our days. Haven't managed to see her just now. Actually I was thinking more in the style of Terminator or Matrix, the machines out clevering us.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...