islamics:

Aah! We are in the same boat ai. Me am learning Android right now. I just do like @SoftEng suggested, I will build my Android skill up, then delve into ML.

I once participated in a challenge and my task was to build a bot using NLP, I wanted using DialogFlow but I was told to build a custom NLP for the nature of my task, na so I give up o.

Lol... We will get there brother

Chukwudaalu:
So good to be here. I suggest we make use of a channel where we can easily share articles, videos, PDF files, pictures, etc. Seasons greetings.

Do you have any platform suggestion in mind?

BTW, we can also share articles, videos etc via links on Nairaland as well.

islamics:

Aah! We are in the same boat ai. Me am learning Android right now. I just do like @SoftEng suggested, I will build my Android skill up, then delve into ML.

I once participated in a challenge and my task was to build a bot using NLP, I wanted using DialogFlow but I was told to build a custom NLP for the nature of my task, na so I give up o.

OK. All the best with your Android study.

SoftEng:


Do you have any platform suggestion in mind?

BTW, we can also share articles, videos etc via links on Nairaland as well.

It's all good then. Let us continue on this platform. We might later have interested persons joining us. Thanks.

Lemme start following here.

SoftEng:


Do you have any platform suggestion in mind?

BTW, we can also share articles, videos etc via links on Nairaland as well.

i suggest we use telegram, because it group can accommodate more members compare to whats, you can share file of bigger size on telegram, and it has a desktop app which whatsapp also have. I have been a member of likw 4 programming groups on whatsapp and 3 on telegram for over 1 year. The members meet on nairaland and come together to create the group on whatsapp, its easier asking questions there than coming to nairaland to ask question. And the members see themselves as family

can someone give a table of content on things i need to learn on A.I and machine language from basics

Let get a WhatsApp grp

I would like to suggest we get the platform issue out of the way b4 moving forward. Op,I would suggest we do a quick vote on which platform to use.
So far we have:
1. Nairaland: I know it's a great forum but I feel it lacks some features required for large collaboration.
2. Telegram: I have never had to use it but I hear it's quite good.
3. WhatsApp: Great app but somehow every group I've been part of died a slow death.
4. Slack: My personal choice.
Of course there are plenty other great platforms that can be suggested.

I truly have high hopes for this group and starting on the right foundation and structures would go a long way in making it sustainable and scalable long term.

Following

omokoladejames:
I would like to suggest we get the platform issue out of the way b4 moving forward. Op,I would suggest we do a quick vote on which platform to use.
So far we have:
1. Nairaland: I know it's a great forum but I feel it lacks some features required for large collaboration.
2. Telegram: I have never had to use it but I hear it's quite good.
3. WhatsApp: Great app but somehow every group I've been part of died a slow death.
4. Slack: My personal choice.
Of course there are plenty other great platforms that can be suggested.

I truly have high hopes for this group and starting on the right foundation and structures would go a long way in making it sustainable and scalable long term.

I agree with you that we have conclude on matters related to platform. Thtat's why I have been asking for suggestions from everyone who showed interest in this group. Eveeryone's suggestion becomes some sort of poll or vote and the platform that has the highest number of suggestion should be used.

However, not many persons have suggested or contributed any comment to discussion centering around platforms. Please everyone should comment. I don't want us to take a decision based on 1 or 2 suggestions only. Thank you.


cc:
Dmayor7
osarenomaspecial
unphilaz
shams040
Samcent
islamics
lum1
kelvine95
4kings
rayval
Ibroh24
BunchObute
NaHiim
EbonyMira
themonk7
Unmarshalled
Omoadeola
lydia4211

@Dell031
I earlier shared the below post. It should give help you extract some points (like a "soft table of content" about topics you need to go through.
Thanks.

SoftEng:


4kings thanks for your input. Suggestions from others are also welcomed.

My Suggestion
In addition to 4kings suggestion, the links below are short blog posts that gives some perspective.
1. https://www.digitalocean.com/community/tutorials/an-introduction-to-machine-learning
(gives a very high level introduction to Machine Learning)

2. https://www.toptal.com/machine-learning/machine-learning-theory-an-introductory-primer
(similar to the first link, but goes a little bit deeper, with very little amount of math to describe the algorithm that powers some methods)


Lastly, if you more interested in getting a well rounded knowledge (not just high level overview), then any
1. If you are interested in first getting a rounded knowledge about Machine Learning, then link below is a good introduction
https://www.coursera.org/learn/machine-learning.
The course is video based, so you will need a fair amount of internet data.

2. If you can't do video (because of data), the website below is also nice, but you will need to browse through it (especially the blog section) to mine useful information from it.

3. The link below (machine learning cheat sheet with much emphasis on neural networks) may also be nice.
https://ml-cheatsheet.readthedocs.io/en/latest/index.html

In my own view: personally, I would prefer Telegram because it is more Developer friendly and the audience could be worldwide. Though WhatsApp has more users. Again, Telegram has API for those that might want to play with it. One of the reason why I think WhatsApp group die easily is that they are yet to allow incorporating of bot into the app (though I don't know if they have) . For instance, in telegram, in some group when you join; you are welcome by a bot. In some cases, a bot provide updates on different materials. And there is a pinned message available to user. Again new user can see previous message in a group unlike WhatsApp.
Any one decided on is welcome. Even though, it is slack.

islamics:
In my own view: personally, I would prefer Telegram because it is more Developer friendly and the audience could be worldwide. Though WhatsApp has more users. Again, Telegram has API for those that might want to play with it. One of the reason why I think WhatsApp group die easily is that they are yet to allow incorporating of bot into the app (though I don't know if they have) . For instance, in telegram, in some group when you join; you are welcome by a bot. In some cases, a bot provide updates on different materials. And there is a pinned message available to user. Again new user can see previous message in a group unlike WhatsApp.
Any one decided on is welcome. Even though, it is slack.

seconded

SoftEng:


I didn't want to comment on this because of my limited knowledge in NLP. The only area(task) that I have dabbled into in the past is Language Modeling. However, from what I remember Question Answering Systems (QAS) are still very much limited as researchers have not been able to get it to practically work well on open ended domain. This does not mean that it doesn't work, but it just very hard right now. It's does not work practically well compared to Language Translation task which works well.

Here's something that maybe useful to you. Amazon hosts an Alexa Prize competition (https://developer.amazon.com/alexaprize) for universities around the world to build a conversational agent (which is atleast similar to QAS, if not thesame). You can check out the 2017 prize winner and read through their paper or blog post (if available) to see the techniques used. Who knows, if they decide to open it to candidates outside universities, you can form a group here that can participate as well.

Below are 2 additional links (not related to the competition) that you might find useful
https://aws.amazon.com/blogs/ai/creating-a-question-and-answer-bot-with-amazon-lex-and-amazon-alexa/
https://aws.amazon.com/blogs/aws/announcing-the-aws-chatbot-challenge-create-conversational-intelligent-chatbots-using-amazon-lex-and-aws-lambda/

[url]Question[/url]
Do you have any dataset in mind that you want to use to train your QAS?

Final Note
I once watched a video where a top NLP scientist mentioned that QAS are like crown jewel of NLP because a fully functional QAS should be able to perform other NLP tasks (e.g. you can ask a QAS to translate a text from one language to another, or you can ask a QAS to summarize text)

Thanks.

The usual method for question answering normally revolves around the popular bag of words model, where faetures are trained using the best fit algorithm or ensemble methods, and other effective stuff come to play.

I've built some naive question answering system in time past using these methods, and POS taggers are only(or mostly) for word-category disambiguation for context meaning.
For example, the sentence; "He fishes fish", it enables the system to distinguish the verb in the statment.

What i need however is a different approach, with focus on POS taggers. I just want to see the extent a good analysis of POS taggers can be used for question answering without ngrams or bag of words model or lemmatization and so on, and because of the nature of POS taggers, i think if one is succesful in closed domain questions, open domain won't be a problem, because the algorithm won't be using a similarity metrics per say, but will actually understand the words spoken.

So this is more of a research to me, when i'm less busy I'll work up some stuff to see what i come up with and share with the group.

I've used POS taggers for speech synthesis, and this was actually the only place where POS taggers really matterd the most.

islamics:
In my own view: personally, I would prefer Telegram because it is more Developer friendly and the audience could be worldwide. Though WhatsApp has more users. Again, Telegram has API for those that might want to play with it. One of the reason why I think WhatsApp group die easily is that they are yet to allow incorporating of bot into the app (though I don't know if they have) . For instance, in telegram, in some group when you join; you are welcome by a bot. In some cases, a bot provide updates on different materials. And there is a pinned message available to user. Again new user can see previous message in a group unlike WhatsApp.
Any one decided on is welcome. Even though, it is slack.

I agree with this, whatsapp is not a good platform for a group like this.
I've never used telegram though, but from this explanation i think Slack or Telegram would be Okay.

Though I'm mostly used to Slack.

I suggest we should just start sharing ideas on this medium(nairaland) since apparently we all have access to it. Its the only platform that I currently have access to for instance. Forgive my ignorance but I've never heard of slack and telegram.

All suggestions made are amazing but I would suggest we maintain this platform for now pending when we have reached a level where we can now form a network of serious minded Airtficial intelligence people because TBT. we haven't even really discussed much here just a few questions and answers and although people are learning but we need to do more and we should start from here before moving so we won't face issues been faced by groups on (W, T, S).

raymod170:
All suggestions made are amazing but I would suggest we maintain this platform for now pending when we have reached a level where we can now form a network of serious minded Airtficial intelligence people because TBT. we haven't even really discussed much here just a few questions and answers and although people are learning but we need to do more and we should start from here before moving so we won't face issues been faced by groups on (W, T, S).

well said bro.

All tech companies are already implementing Ml and AI into their product and services how can we do something like this here in Africa or Nigeria because we all know the amount of data they get from us which they use to carry out these experiments before selling it as an improved product or service back to us and we crave for it badly

watch this google i/o

https://www.youtube.com/watch?v=CNLVZjBE08g

how do we create this here in Africa to help improve out selves cos the bitter truth we Africans don't even have this much data not to talk of designing systems that improves our day to day activities

data is key and I feel tho we should be careful with the data we have but at the same time start small and intergate Ml and AI into the apps we build or websites we develop with the available data at our disposal then gradually we would then look to make it bigger

but the question remains how can we do this cos I myself am in my room on my laptop making research and thinking of how to implement them on my softwares and websites going forward

best regards

Nice thread........impressed.
Just about to start learning javascript today (intend to spend the next 2-3 doing an intensive learning). My aim is to become SAP HANA XS and SAPUI5 consultant before end of 1st quarter next year. After javascript, AI and ML is the next level ( SAP Leanardo).
Will definitely follow this thread...........

I am who I am today because of Nairaland.........I got to know about SAP 11years ago through Nairaland.

Kudos to you all

4kings:

Thanks.

The usual method for question answering normally revolves around the popular bag of words model, where faetures are trained using the best fit algorithm or ensemble methods, and other effective stuff come to play.

I've built some naive question answering system in time past using these methods, and POS taggers are only(or mostly) for word-category disambiguation for context meaning.
For example, the sentence; "He fishes fish", it enables the system to distinguish the verb in the statment.

What i need however is a different approach, with focus on POS taggers. I just want to see the extent a good analysis of POS taggers can be used for question answering without ngrams or bag of words model or lemmatization and so on, and because of the nature of POS taggers, i think if one is succesful in closed domain questions, open domain won't be a problem, because the algorithm won't be using a similarity metrics per say, but will actually understand the words spoken.

So this is more of a research to me, when i'm less busy I'll work up some stuff to see what i come up with and share with the group.

I've used POS taggers for speech synthesis, and this was actually the only place where POS taggers really matterd the most.

Alright. Good luck.

You should also checkout http://www.maluuba.com/research/ as suggested by osarenomaspecial

Hi all,

Thanks to everyone who gave their opinion with regards to which platform (e.g. Nairaland, Telegram, Whatsapp, Slack etc) the group should use.

I conclude that we should keep discussion here on Nairaland. No need to migrate elsewhere. The main benefit I envisage in staying on Nairaland is openness. Openness, meaning that our discussions can be viewed by the public (both to interested and non-interested persons) and the possiblity of this thread popping up on search engine results for anyone who search for Artificial Intelligence (AI) and Machine Learning (ML) community or group in Nigeria.

Thanks all. Looking forward to all the ideas/research/projects/discussions that will unfold over time.

Please can someone help me on bow to download and install python, d one I downloaded is telling me error 0x80240017.

SoftEng:


Alright. Good luck.

Thanks

You should also checkout http://www.maluuba.com/research/ as suggested by osarenomaspecial

Sure.

odizeey:
Please can someone help me on bow to download and install python, d one I downloaded is telling me error 0x80240017.

That's windows wahala.
Download Python 3.4

4kings:
That's windows wahala. Download Python 3.4

thanks boss

Please can anyone give me dataset to work on. To practice thanks.

odizeey:
Please can anyone give me dataset to work on. To practice thanks.

Register on Kaggle

Check the list of competitions you can practise from, you can start with the popular titanic disaster challenge

They're many datasets to practise on Kaggle, I'll actively be participating there also next year.

4kings:

Register on Kaggle

Check the list of competitions you can practise from, you can start with the popular titanic disaster challenge

They're many datasets to practise on Kaggle, I'll actively be participating there also next year.

thanks boss.

SoftEng:



I agree with you that we have conclude on matters related to platform. Thtat's why I have been asking for suggestions from everyone who showed interest in this group. Eveeryone's suggestion becomes some sort of poll or vote and the platform that has the highest number of suggestion should be used.

However, not many persons have suggested or contributed any comment to discussion centering around platforms. Please everyone should comment. I don't want us to take a decision based on 1 or 2 suggestions only. Thank you.


cc:
Dmayor7
osarenomaspecial
unphilaz
shams040
Samcent
islamics
lum1
kelvine95
4kings
rayval
Ibroh24
BunchObute
NaHiim
EbonyMira
themonk7
Unmarshalled
Omoadeola
lydia4211

Lets use slack or telegram