machinephile

programming, maths & philosophy that binds all together (ocational rants)

[just-desparate-machine]

In computing various relations in high-dimensional metrics (which is a common practice in machine learning, and other computational applications), Tensors come in handy.

Tensors may sometimes be hard to understand as it is very abstract and has variable intuition depending on the context, for example understanding tensors for physics (understanding big things in the cosmos) or abstract mathematics (to understand logical bindings). That’s why there may be many different explanations. But for now, we are going with bases in high-dimentional vector-spaces 

A general explanation of Tensors in liner algebra for computation goes as follows; this explanation is attributed from one great answer I found on the mathematics stackexchage.    

Found it helpful, let me know!

[just-desparate-machine]

[just-desparate-machine]

COMMON-SENSE reasoning AI (KagNet)

What type of algorithmic approach does one need to do in order for machine-intelligence to be aware of Common sense knowledge. Unlike other fields in AI, Common sense is subjective where biases and prospectives are variables. What type of flow is necessary to compute common sense?

Comes in this context the Knowledge-aware-graph based convulsional neural network tuned by BERT; conceptnet. The object here is subjected to questions with a certain pair of answer(s) in the context of common sense.

given a question like “Where do

adults use glue sticks?”, with the answer choices

being {classroom(✗), office (✓), desk drawer (✗)},

Thus if, for example an arbitrary question is presented, the question concept would be a function of potential answer-set. Our job is to tune graph accuracy.

Thus,

Specifically, for each question concept ci ∈ Cq

and answer concept cj ∈ Ca, we can efficiently

find paths between them that are shorter than k

concepts4

. Then, we add edges, if any, between

the concept pairs within Cq or Ca

[just-desparate-machine]

[just-desparate-machine]

How to auto-tune Image filters using AI and Computer Vision

[just-desparate-machine]

Zero(close to) data to Generating Complex patterns using Machine Learning? Challenge possible?

Usually, when it comes to training a machine to learn complex concpets, all the data involved in the input field is very large and might also need to be carefully parsed to fit in with the execution schema. And a #datascientist, you might get a 300 pound block on yours chest when you have to #parse huge #datasets. Minimizing the input data examples would be an extraodinary miilestone for machinelearning and will also help you keep your stress levels low. Seems interesting, I bet the following posts may appeal more, Follow along 💗

[just-desparate-machine]

What is Virtualization in Cloud Computing?

[just-desparate-machine]

Web data visualization using Python (libraries to use)  

It’s always fun to find meaning in data. Which’s a better resource to capture data than the vast web we all have access to. Web scraping is pretty useful and fun and visualizing those data is even more fun (what else?! we are visual creatures) Use this combination python liberies to do both() ENJOYED THIS POST? I will have more coming that you will absolutely love, why not follow

[just-desparate-machine]

What is a Headless Browser and why it is important❓

In a world filled with a lot of different web apps, and where your whole infrastructure is on the web, a lot of time and money get spent into managing and operating repeating tasks and this just makes high outputs so more expensive. 

For businesses to stay relevant in their niche marketplace, one of the effective strategies is to spread your digital arms like a net across the vast network in your specific niche and keep track of what's happening all across. 

To accomplish this, let's try out the following: 🚀

Ditch the head; which is the graphical interface for users to interact with the browser and keep all its functionalities. 

How do you tell the browser what to do, and when to do? By writing some neat code. 

Now have the headless browser and the code running on a remote virtual server and boom! You have saved a considerable amount of time and cost. 

As there is no UI being run(head), it can run instances of multiple tasks simultaneously all in the background. The possibilities are big!

[just-desparate-machine]

Pro tip💡: how to use uncle Google like a pro. 

 Along with #google, many search engines follow similar syntax formatting as database query languages such as SQL. When you want to grab very specific information from the wide web (that you won't find otherwise), use this neat cheatsheet ☝️ 

The characters other than text are 'operators', using them you can search very specific types of information from specific sources. These give you more dynamic control over your search, so keep it handy.

[just-desparate-machine]

test

[just-desparate-machine]

Certainly is a great tool for you designers out there. I happen to come across and loved it asa labded.

Make sure to follow my twitter for cool digital info and resources related to dev and computer sci

[just-desparate-machine]

Google's data set search engine is out of beta now. Currently indexing over 25 million trees of datasets in various fields. Most of the instances in these datasets are labeled and all are open-source.

You can also submit your own dataset to its database. This will let almost anybody train datasets with ML algorithms to process, label data for various purposes.