Advantages and Disadvantages of

Machine Learning Language

Every coin has two faces, each face has its own property and features. It’s time
to uncover the faces of ML. A very powerful tool that holds the potential to
revolutionize the way things work.
Advantages of Machine learning
1. Easily identifies trends and patterns
Machine Learning can review large volumes of data and discover specific
trends and patterns that would not be apparent to humans. For instance, for
an e-commerce website like Amazon, it serves to understand the browsing
behaviors and purchase histories of its users to help cater to the right
products, deals, and reminders relevant to them. It uses the results to reveal
relevant advertisements to them.

2. No human intervention needed

(automation)
With ML, you don’t need to babysit your project every step of the way. Since it
means giving machines the ability to learn, it lets them make predictions and
also improve the algorithms on their own. A common example of this is anti-
virus softwares; they learn to filter new threats as they are recognized. ML is
also good at recognizing spam.

3. Continuous Improvement
As ML algorithms gain experience, they keep improving in accuracy and
efficiency. This lets them make better decisions. Say you need to make a
weather forecast model. As the amount of data you have keeps growing, your
algorithms learn to make more accurate predictions faster.

4. Handling multi-dimensional and multi-

variety data
Machine Learning algorithms are good at handling data that are multi-
dimensional and multi-variety, and they can do this in dynamic or uncertain
environments.

5. Wide Applications
You could be an e-tailer or a healthcare provider and make ML work for you.
Where it does apply, it holds the capability to help deliver a much more
personal experience to customers while also targeting the right customers.
Disadvantages of Machine Learning
With all those advantages to its powerfulness and popularity, Machine
Learning isn’t perfect. The following factors serve to limit it:

1. Data Acquisition
Machine Learning requires massive data sets to train on, and these should be
inclusive/unbiased, and of good quality. There can also be times where they
must wait for new data to be generated.

2. Time and Resources

ML needs enough time to let the algorithms learn and develop enough to fulfill
their purpose with a considerable amount of accuracy and relevancy. It also
needs massive resources to function. This can mean additional requirements
of computer power for you.

3. Interpretation of Results
Another major challenge is the ability to accurately interpret results generated
by the algorithms. You must also carefully choose the algorithms for your
purpose.

4. High error-susceptibility
Machine Learning is autonomous but highly susceptible to errors. Suppose
you train an algorithm with data sets small enough to not be inclusive. You
end up with biased predictions coming from a biased training set. This leads to
irrelevant advertisements being displayed to customers. In the case of ML,
such blunders can set off a chain of errors that can go undetected for long
periods of time. And when they do get noticed, it takes quite some time to
recognize the source of the issue, and even longer to correct it.
Artificial intelligence (AI) embodies all the other fields that aim to create intelligent agents.
Though AI is itself embodied by artificial general intelligence or AGI. AI is about making
something to exhibit some level of intelligence such as learning from examples is a good
exhibition of intelligence.

• Machine learning (ML) is more about learning and nothing else. It is about a system with a
poor initial state that incrementally improves itself from examples by altering its state in a
manner that increases its likelihood of becoming better. The system state is determined by its
adjustable parameters and the direction in which to alter the parameters to improve the system is
normally determined by the first order gradients. Machine learning is a fine tuning process that
slowly improves a system. Typical ML algorithms are based on gradient decent such as
stochastic gradient decent algorithm.

• Genetic algorithms (GA) are under ML because they are also about learning, that is,
improving a system incrementally that initially is in a bad state. Though GA is based upon
natural selection. Several best systems among many others get to have the chance to pass their
characteristics to the next generation. These characteristics are actually parameter settings, thus
the GA algorithm can pick best k systems with the best parameter settings and determine the next
generation by combining these parameters in a particular or random manner.

• Artificial neural networks (ANN) are under ML as well because they are learning algorithms
that incrementally get better from examples. ANNs are an ensemble of processing nodes
arranged layer-by-layer. They normally represent data in a distributed hierarchical manner. The
first layer learns simple atomic representations whereas the upper layers learn more complex
abstract representations. Typical ANNs are the old neocognitron and multi-layer perceptrons
with few layers deep.

• Deep learning (DL) is under ANN and DL systems are ANNs on steroids, that is, they are
very large with many more layers than a typical ANN. Normally the DL term is taking over the
ANN term in most cases. A typical DL system is the convolutional neural network which can be
30+ layers deep.
At a high level, AI tries to create general purpose intelligent things, kind of like a human. ML
is usually used for this purpose, though ML tends to be used to solve specific problems that
it was trained to solve. Artificial Neural Networks are a subset of models that are used in
Machine Learning, they are characterized by their use of neurons. Deep Learning is a subset
of ANNs that are characterized by their use of multiple hidden layers of neurons.

Artificial Intelligence

AI is a field that tries to create general purpose models that are “intelligent” by some
definition of that word. Generally, AI models try to solve a huge variety of models, even
some which the model may not have been built with the idea to solve. Commonly AI is built
on top of lots of ML models working together, where an ML model is kind of like specific AI
as opposed to general AI. It’s usually trained to solve a single problem.

Machine Learning

Machine Learning is creating a function that mimics the patterns in data. For example, if I
have a dataset indicating age, education level (# 0–10), average salary in location, and
annual income. I can use machine learning to generate a function that guesses annual
income given age, education, and location. A simple way to do this, could be to create 4
weights A, B, C, and D and make:

annual income = A * age + B * education-level + C * average-salary-in-location + D

Now at first A, B, C, and D will be initialized randomly and using the data A, B, C, and D will
be tweaked until we get good results.

The format of the function is the model. There are many different models that can be used.
This model in particular is a linear model, and assumes that annual income
increases/decreases linearly with age, education level and average salary in location. If this is
false, and the relationship is actually non-linear, you’ll get bad results. This is why the choice
of model is important.

Artificial Neural Network

Is another model, like the linear one mentioned above, except it has the form:
It is made up of layers, where each layer is densely connected to the next layer. he inputs
are provided in the first layer, in this case it’d be education level, age, and average salary in
location. The output would be a single node which would indicate the model’s prediction of
salary. The hidden layer is used to attempt to model the function. Each node combines the
inputs from the previous layer and uses an activation function and weight to decide what
value to send to the next node. The weight of each edge is the part that gets tweaked for
better results.

As opposed to the model mentioned before, neural nets can model non-linear functions
making it better for many problem. However, the results are harder to understand and
neural nets generally take much longer to train. A neural net also requires you to manually
chose number of nodes and number of layers.

Deep Learning

Deep learning is a bit of a buzz word. It’s any Artificial Neural Network that has more than
one layer. It’s been proven that any function that can be modeled by multiple layers can also
be modeled by a single layer; however, in practice deep learning often results in better
results with less nodes, faster. There are also other types of neural networks like Recursive
Neural Networks that feed input from one node back into itself to deal with things like time-
series data. This is also a part of deep learning.
Artificial Neural Network (ANN)

A Single Neuron inside Brain

A Single Neuron viewed through ANN model

A Typical Brain Image with respect to Neuron’s interconnection
 Representation of a single Neuron and ANN Node

Similarity between Biological Neural Network Artificial Neural Network

ANN goes back as early as the 1950s, an example being the creation of
perceptrons. The idea behind perceptrons was to emulate what we thought the
BNN does in our brains. In the biological neural network, dendrites receive signals
from other cell bodies which they use the nucleus to process and send out to others
with the help of axons. The link between one nucleus and others can vary, leading to
the storage of information (i.e. memory) and the creation of new neural links. What
happens during the processing stage in the nucleus still remains uncertain, however,
the ANN also tries its best to mimic the workings. The ANN also receives inputs and
uses its hidden layer to process information and send out the results. However,
unlike the biological neural network, the ANN is a mathematical model with
changing parameters and functions to calculate whatever it is that we require of it.

Biological Neuron vs. Artificial Neuron

The biological neuron is analogous to artificial neurons in the following terms:

Biological Neuron Artificial Neuron

Cell Nucleus (Soma) Node

Dendrites Input

Synapse Weights or interconnections

Axon Output

Differences:

We have over 86 billion neurons in our brain creating and over 100 trillion synapses
(links) while ANN consists of a number much smaller. In terms of speed of
operations, the neurons in a human brain can vary widely due to aspects such as
age, gender, how much they got the night before, etc. It is much easier for the ANN
to stay consistent in its calculations. However, it is important to understand that
consistency comes from the specialization of ANN. An example being an ANN
designed to play chess wouldn’t be able to play checkers unless designed to.

Types of Artificial Neural Networks

Generally, there are two types of ANN. Such as FeedForward and Feedback.

a. FeedForward ANN
In this network flow of information is unidirectional. A unit used to send
information to another unit that does not receive any information. Also, no
feedback loops are present in this. Although, used in recognition of a pattern.
As they contain fixed inputs and outputs.
. FeedBack ANN
In this particular Artificial Neural Network, it allows feedback loops. Also,
used in content addressable memories.
Artificial Neural Networks Applications
Artificial Neural Network used to perform a various task. Also, this task
performs that are busy with humans but difficult for a machine.
a. Aerospace
Generally, we use ANN a for Autopilot aircrafts. They used for aircraft fault
detection.
b. Military
In various ways, we use ANN an in the military. Such as Weapon orientation
and steering, target tracking.
c. Electronics
Basically, we use an Artificial neural network in electronics in many ways.
That are code sequence prediction, IC chip layout, and chip failure analysis.
d. Medical
As medical has too many machines. That use in various ways. Such as cancer
cell analysis, EEG and ECG analysis.
e. Speech
We use ANN in speech recognition and speech classification.
f. Telecommunications
Generally, it has different applications. Thus, we use an Artificial neural
network in many ways. Such as image and data compression, automated
information services.
g. Transportation
Generally, we use an Artificial neural network in transportation in many ways.
That are truck Brake system diagnosis and vehicle scheduling, routing
systems.
h. Software
It also uses an ANN in pattern Recognition. Such as in facial recognition,
optical character recognition, etc.
i. Time Series Prediction
We use an Artificial neural network to predict time. Also, we use ANNs to
make predictions on stocks and natural calamities.
 Perceptron

The perceptron is the building block of artificial neural networks,

it is a simplified model of the biological neurons in our brain. A
perceptron is the simplest neural network, one that is comprised
of just one neuron. The perceptron algorithm was invented in 1958
by Frank Rosenblatt.
The perceptron is a simplified model of the real neuron that
attempts to imitate it by the following process: it takes the input
signals, let’s call them x1, x2, …, xn, computes a weighted sum z of
those inputs, then passes it through a threshold function ϕ and
outputs the result.

But having w0 as a threshold is the same thing as adding w0 to the sum as bias and having
instead a threshold of 0. That is, we consider an additional input signal x0 that is always set to
1.