Technology disrupts many
jobs but it creates new jobs with more efficiency and consistency to become the
lifeline for economic growth of any country”.
Machine Learning (ML) is a subset of AI and the most popular technology of the 21st century. We all know that traditional automation has one input to a computer resident program which processes input and gives output for further human analysis and interpretation. However, ML is an iterative process, which gets one input and its expected output, both acting as further inputs to the computer where software and algorithm are resident. It iteratively processes both inputs and generates a program that produces a final outcome. ML is a very laborious process to cater for all types of constraints and make the computer act in an autonomous mode but ensuring that its outcome is timely and trustworthy for decision making.
Development of ML system is
like software project which requires many experts of multi-disciplines to
collectively evolve required algorithms. ML system carries out exhaustive
numerical processing, analysis and interpretation for making good decisions.
However, ML is not just software programming but the very iterative process of
training a computing machine to respond
like a human expert to emerging situations. It is like a mother teaching her
small child to store various options in his/her brain and when required, respond
very intelligently. A near similar concept is being followed for implementing
ML by leading manufacturers of cars for faults diagnostic and analysis, medical
labs for testing and diagnostic of the disease and prediction say of weather.
The
evolution of Big Data, supported by the enormous computing power of Cloud Computing
has given tremendous momentum to business decision-making. The volume and speed
of incoming data continue to rise rapidly. Consequently, the ability of computing
machines to learn and support intelligent decision-making has become a necessity.
Although, ML may identify previously unidentified problems to be solved, yet
the machine is not autonomously creative and decision-maker. The output of ML
is mainly dependent on the data and algorithm and any change in data will
change the outcome. It is well known that even the best algorithms can't think,
feel, and present information like human experts. ML can just perform
predictive analysis, much faster than any human. Although ML has no fatigue
factor-like human beings and can easily do multi-tasking and work more
consistently and efficiently, yet it cannot fully replace human beings.
Scope of ML. ML is already
being used in many manufacturing industries, hospitals, and research
institutes, insurance sector and legal cases. Consequently, it is creating a
big demand for AI/ML skilled professionals. According to one survey, the ML market
is expected to grow to $8.81 billion by 2022. ML is of great help in researching old data
of viruses and determining the trend of viral spread. AI and ML are already employed in various videogames, entertainment,
surveillance systems, weapon systems and many other applications of space
exploration. In this chapter, the scope is limited to basic concepts of ML and
its applications which help productivity, efficiency, accuracy and job
creation.
ML
Process. Machine (Computer) learns by iteratively processing data to detect
patterns of data and by applying pre-stored decision-making rules. It
faithfully and consistently carries out the following tasks:
•
Categorize.
To categorise customers, vendors, materials, financial potential and
shopping habits, lifestyle, likes/ dislikes of people.
•
Identify.
To identify unknown patterns and relationships among peoples, machines/devices
to forecast emerging trends /risks.
•
Detect.
To detect unexpected behaviour of people or malfunctioning of machines/devices.
•
Predict.
To predict likely outcomes based on identified patterns, business
practices and statistical rules.
•
Scheduling. To
predict waiting/processing time at a facility say for ICU occupancy
in hospitals and carry out schedule.
Algorithm.
The heart of the ML system is the embedded
software (Algorithm) which is developed and validated by a multi-disciplinary
team of experts. Algorithms are set of formulas or equations or matrices or
mathematical expressions, indicating various constraints and range of values
with acceptable tolerance or error. These mathematical equations are
numerically solved by very high-speed processors, data is analysed and the outcome
is made to go through a number of iterations until it is close to the expected
result. Different algorithms process and learn in different ways. As more data
regarding observed responses or changes to the environment are provided to the
computing machine the algorithm’s performance improves. Algorithm development is like a child learning from his/her mother
and improving progressively.
ML Validation Criteria.
ML development requires a very close interaction between a subject expert and
the computing machine. ML requires the application of the scientific method and human communication skills. Human
knowledge and machine must complement each other. ML by itself can predict
emerging trends as per algorithm but cannot give the final conclusion. The
process of selecting, auditing and fine-tuning an algorithm to deliver reliable
results is very important to ML applications. Selection of the
appropriate algorithm is crucial for problem-solving and data under
investigation. For validation of algorithms following factors should be considered:
·
Hypothesis. The design team must know what they are trying to predict. Therefore ML
project should start with a clear statement of the problem and the hypothesis
to be investigated.
·
Test
Features. Data scientists and subject matter experts must work together to find
out the sources of data and the key features of the computing machine. Data
visualization can play a key role in helping to highlight and testing features
that can be fed into ML algorithms.
·
Pre-
Processing. Check if the data be pre-processed, cleaned and, transformed to
facilitate further processing and better results.
·
Data
Features. Identify what data features should be included and excluded in
formulating the algorithm.
·
Optimisation. Find
out how algorithm’s parameters can be
fine-tuned for optimal performance
·
Cause and Effect. Be
sure if the resulting correlations are predictive or causal.
·
Exceptions.. Be
clear about exceptional conditions and their
acceptable limits
Indian National Strategy on ML. In 2015 India had taken a very bold imitative on “Digital India”. To give further impetus to its digital economy, NITI Aayog (Planning Commission) of India had formulated comprehensive ( 28 Pages) guidelines in Jun 2018, for introducing AI, ML and Robotics in various sectors like Agriculture, Healthcare Weather Prediction,, Industrial Automation, Research & Development., Gaming and Entertainment, Law Enforcement, The details of this strategy are available on the web. ww.startegy.org.in
Applicability of ML. Today, the corporate world is better equipped with Market Intelligence
(MI) / Business Intelligence ( BI) like customers’ buying capacity, frequency,
preferences and lifestyle. This MI/BI comes from fast analysis of shopping data
and social media information as to why customers of a particular country,
community, gender, age and financial status buy their products, use their
services, or engage their expertise. Earlier, such activity of MI/BI has
been done through the company’s sale persons. market survey, business experts and
in-house IT resources. With today data deluge in social media, data is really
huge in terms of its Velocity, Variety and Volume and Varsity (V4).
Its fast processing and accurate
analysis need assistance from very fast computing machines. With enormous data storage
and processing power available through Cloud Computing, the computing machines can
quickly detect purchase patterns and preferred channels partners to support the
sale of their products. ML is particularly suited to problems where -
·
Rules/constraints cannot be easily codified by simple logical rules,
·
Accuracy is more important than interpretation or interpretability,
·
The data is too huge and complex for traditional analytic techniques.
Job Potential of ML. Most of the medium to large industries dealing with large amounts of data have recognized the
value of ML technology. ML facilitates online monitoring through embedded
sensors, real-time processing and timely generating output for human
decision-making.. The range of ML applications depends upon human ingenuity to
evolve good algorithms and the ability to interpret ML outcomes. Some of the common
areas which have good job potential are briefly given below.
· Healthcare Services.
§ Monitoring patient health. The use of ML is fast-growing, particularly in healthcare services.
With the advent of wearable devices and smart sensors, a patient's health can be
monitored in real-time.
§ ML system can also monitor the availability of beds in ICU and facilitate scheduling of waiting patients..
§ Tele Medicine, There is an acute shortage of doctors and
nurses in Indian hospitals. India has also an issue of long distances between
its Primary Health Centers (PHCs) located in far-flung areas and referral
hospitals located say 200Kms -1000Kms away in cosmopolitan cities.
Tele-medicine using ML technology could mitigate this distance problem.
§ ML can also help to identify trends in serious diseases and that may
lead to improved diagnoses and treatment. It can also be of great value
in combating Corona like pandemic
§ ML integration with other
technologies. ML combined with the Internet of Medical Things (IoMT) and
Drones will be the new mantra for planning and providing quality healthcare at
affordable cost and to a greater satisfaction to the patients.
·
Genetic analysis. ML
Based on the patient’s genetic information and, demographic information can
also determine which treatment will be most effective for an individual
patient. Every human being has more than 20,000 genes. ML can easily handle
integrated data from multiple, sources. Earlier scientists were using
super-computers for analysing patient data related to genetic disorders for
various diseases. They could trace back say 2-3 generations. More often past
records were not available for drawing any co-relation. ML
technology has opened a new horizon for genetic analysis
and interpretation.
·
Digital Manufacturing. In
highly developed countries, manufacturers of various products are applying ML
techniques to identify likely equipment/ component failures, before those
actually fail. This requires fitting appropriate sensors in various vehicles,
machines, equipment, appliances and assemblies, enabling those to automatically
communicate quantified parameters to the central database. It will provide great
satisfaction to the customers in terms of service, maintenance and warranty policies. ML will enable sharing of
product-related information between the design team and the manufacturing team. ML can
greatly reduce design effort, production time and overall cost. It will also
provide efficient documentation for version control and easy traceability of
defects.
·
e-Governance. India has taken many
initiatives to switch over from traditional paper-based administration to e-Governance. This has
really boosted the Indian economy and global participation. Government agencies like land records, demographic data, health data,
Road /Rail/Port data, public safety and utilities have a particular need for
ML since they have multiple sources of data that need fast handling and policy
action. Data analysis by ML method helps
to identify ways to increase the efficiency of their services and save funds. ML
can also help detect fraud in digital transactions.
·
Banking Services. ML has major
applications in banks for approving loans to individuals, industries, farmers,
Research Labs and other financial institutes, ML can help to detect and prevent
frauds.
·
Financial Investments. ML can monitor market trends,
and offer profitable investment opportunities, to their registered investors.
ML can also identify potential clients with high-risk profiles, and
invite them to invest.
·
Retail Sale. With advances in various multi-media Apps, the company portal/Website can recommend items you might like to buy, based
on your previous purchases and wish-list. Retail stores/malls are using ML to
analyze your buying history and financial capacity. They also use ML to
implement a marketing campaign, price setting, planning
product delivery and for Customer Relation Management ( CRM).
·
Oil and Gas Exploration.. ML helps in locating new oil sources both off-shore and on land. ML
can help in analyzing minerals in the ground. By streamlining oil pipeline
data, ML can make oil transportation and distribution more efficient and
cost-effective.
·
Transportation. Identifying patterns and trends in transporting various goods is the
key to the operations of the transportation industry. These industries need timely
and accurate information about problems on various routes. ML can help in suggesting
drivers about routes that are more safe and efficient for reaching their
destinations.
·
Driving assistance. Unconsciously,
we all use the Dynamic Programming Technique of Operations Research (OR) while
navigating our car /bus/truck through busy routes to reach from point A to D.
Today, GPS keeps giving us near real-time traffic status about
road conditions, say traffic jam on a bridge, and road-block due to some
vehicle accident, landslide or road under repair. This helps
the driver to slow down and quickly decide either to wait there or switch
to an alternate route to reach his/her destination. We do this action on the fly and intuitively make quick decisions to select our route and carry on.
With the availability of new technologies, ML can identify the best routes from
point A to D, consider road conditions, weather conditions, travel time and
predict the best route based on current location and emerging road conditions.
·
Driverless Car.
After Google had launched its driverless cars in mid-2017 in California, USA,
many such intelligent cars /buses have come on the road in a few more countries.
ML can drive a car without requiring input from a driver. This is a great
boon for senior citizens or handicapped persons or those who cannot drive a car
effectively. Vehicle driving actions can
be to start, stop, go forward, go backward, turn, speed up and slow down.
However, the decision to take any such action is influenced by many factors
like road conditions, weather conditions, presence and behaviour of other
vehicles. While a human driver instinctually assesses all these inputs quickly,
formulating rules and embedding those as an algorithm in a vehicle is quite
complex. This feature is being successfully incorporated in new series of cars
which will delight senior citizens.
·
Deep Learning (DL). DL is an advanced ML
technique that makes use of neural
networks. However, data models generated
in DL are more complex than traditional Neural Networks. It follows the concept
of Reinforcement Learning (RL) which is equivalent to teaching someone to play a particular game. The
rules and objectives of the game are clearly defined. However, the outcome of
any two-sided game depends on the judgment of the player who must adjust
his/her approach in response to the skills and actions of the opponent. This is
a responsive or dynamic game where the outcome of every situation acts as an
input to the next situation while the game is being played by two opposing
persons or two teams. DL engages many ML systems and has further improved our
ability to understand analyse and interpret an image, sound and video. This has
been made possible by various advances in ML research and the availability of very
large data and very fast computing machines.
·
On-line product sale.
ML can be used for real-time online pricing and selling products/services. In this case, the computing machine determines the optimal price while humans validate and decide the final sale price. A feedback
loop allows the algorithm to learn from actual sales results. In Amazon’s
automated distribution centres, humans do the packing, while robotic systems
collect required supplies and ensure that the right stuff gets in the right box.
This requires a close interaction
between human and robots.
·
Insurance. In many countries, insurance
promotion activities are being done by middlemen (insurance brokers or
insurance agents). These can be done now more efficiently by using Big Data, AI
and ML. Such new analytic tools will automate determining the total amount a
person is qualified for insurance and for what period. It will also compute the premium to be paid and the terms of payment. Hence the process for issuing an
insurance policy or clearing various claims/bills will be speeded up.
·
Journalism. Many routine jobs of journalists, press reporters and editors can now be
automated using AI and ML tools. Initially, machine capability will be used in
reporting of events like road conditions, traffic jams, weather reporting,
which rely heavily on data and numbers. As more people will use smartphones and
wearable devices, they will get timely, accurate and consistent information ML
integration with Drones will be the game changer in news reporting. This will reduce the number of press reporters moving around in villages and towns for
collecting information about various events.
·
Financial industry. In most banks,
tellers have already been replaced by ATMs and very soon even loan officers and
account managers could be easily replaced by automated systems. Software
algorithms embedded in ML systems can quickly analyse financial data and prepare
accounts statements or balance sheets without the need for accountants. Each
individual can easily prepare his/her Income Tax statement and submit his/her
ITR online without paying a fee to the Charter Accountant (CA). Even Income Tax
departments of the government are now using Big Data, AI and ML to check tax
returns of corporate as well as those of individuals and identify potential
fraud cases.
·
Search Legal Records to Assist
Lawyers. During the initial phase of any lawsuit, lawyers and Para-legal staff
have to sift through many records and books, depending on the complexity of the
case. The use of Big Data and ML can speed
up sifting through many legal documents. The introduction of software like “Discovery”
capable of analysing large volumes of legal documents is expected to reduce the
number of legal clerks and paralegals, As
ML application matures, it will be
possible to train the computing machine to review past legal documents, precedents, past decisions
on similar /near similar cases. Thus, the ML system can even draft a legal brief
for the senior advocate who is handling the case in the court. A statistical
model using ML could even predict the outcomes of major legal cases and make
many highly paid legal experts redundant.
·
Fraud
Detection. All banking systems are already using AI to detect any fraud in using our
credit cards. We all get SMS messages from our credit card bank, checking whether we made a particular transaction using
our credit card. The bank is alerting us that someone else could be doing
shopping using our credit card. ML is a step further to avoid misuse of crest cards and assure cardholders of the safety and
security of their financial transactions.
·
Sea Animal protection: Every year, many sea animals get hit by ships. ML algorithm could guide ships to avoid sea animals by learning the
sounds and characteristics of the animals and the route/location of ships.
·
Predicting waiting times: Most people don't like waiting in a queue. Quite
often, they have no idea as to how long they have to wait. ML helps to
determine waiting times based on staffing levels, staffing load and availability
of resources.
Issues in implementing
ML.
·
Lack of standards. Although
AI and ML have both been around long enough yet there are no specifications or,
standards for either of these.
This makes some people unsure of the outcome of ML-based jobs.
·
Insecurity of Jobs.
Some less informed people fear that AI and ML both are job snatchers and
therefore they avoid/delay implementation of M. This is not really true
as both AI and ML offer new job opportunities, which of course need new
skills.
·
Lack of trust in the
outcome. As we know.it is the algorithm
in the ML system which determines how to process big data and
interpret its outcome. We know that algorithms process input data in specific
ways and create predictable outputs, based on the data patterns. Initially, people are not sure of ML outcomes. However,
in the last 3-4 years lots of confidence has been built-in ML.
Summary. Rapid developments in
technology and increased use of AI, Robotics, Drones and ML have enhanced the
accuracy, productivity and quality of various products/services. To deliver trustworthy results, ML algorithms
must be continuously refined, based on data that reflect the current situation.
This is true whether evaluating the impact of new customers or retention of
older ones or rebalancing the power supply network to account for unexpected
spikes in energy demands and avoid blackouts. ML relies on programming
languages like Python, Ruby, Scala and R. Computer Professionals who
implement algorithms should use a programming language that suits best for the project/
task.
If you apply the ML model diligently,
you can determine the customer behaviour, lifestyle, buying capacity, shopping
habits and frequency of shopping. ML can pre-process a lot of data and predict
the outcome for the subject experts/professionals to utilise for further decision-making.
This ML outcome speeds up their decision
making. With help of ML you could be encouraging consumers to buy more, utilize
different commercial channels, user-friendly navigation and driving assistance.
For successful implementation of ML, it is very necessary to keep all
interconnected devices duly updated/upgraded. Likewise, various algorithms,
hardware, embedded software, application software and data interfaces should be
kept updated. Needless to say that ML offers the most lucrative and highly paid
jobs for which you must quickly learn new technologies, be proficient and
become highly productive and billable.
Comments