Machine learning being used to detect fraud
By feeding data into program, cases can be flagged for investigation
Singapore is using machine learning to help detect crime.
SkillsFuture Singapore (SSG), an agency that gives out grants and funding for lifelong training, announced yesterday it has implemented the use of machine learning to detect fraud.
The move is one of SSG's new enhancements to its fraud risk management system.
The media release outlining the enhancements follows several cases of fraud and cheating involving the schemes, including news of a man jailed on Tuesday for his role in a $40 million scam - the biggest fraud case involving a public institution in Singapore history.
The enhancements were recommended by a task force formed about a year ago to conduct a review of SSG's grant administration and fraud mitigation capabilities.
The first enhancement is the setting up of a fraud and enforcement division, allowing for focused investigation, enforcement and monitoring and detecting of suspicious claims.
The second is the use of machine learning as part of a fraud detection system.
Typically, machine learning requires the input of data, allowing for the machine program to identify patterns.
As SSG feeds the program data of previous cases of fraud, cases with similar patterns can be detected by the machine and flagged for investigation.
The release said SSG will continue to feed data of newly confirmed cases to the machine, allowing it to refine its own detection prowess over time.
The third enhancement is the adoption of a range of enforcement strategies when handling cases.
During investigations, immediate administrative actions will be taken to withhold disbursement. Funding will also be immediately terminated when a case of suspected fraud or abuse is established.
News of the largest case of a public institution here being defrauded made headlines this week.
Lee Chi Wai, 32, who used to be a trainer with an insurance firm, stashed $6.7 million in cash and 11kg of gold valued at $626,500 in his Sengkang flat.
His sister and brother-in-law were allegedly pulling off the $40 million SkillsFuture scam and needed a place to park their new-found wealth.
Between May and October last year, nine shell companies were allegedly set up and used to receive $39.9 million from more than 8,000 fraudulent course fee applications.
After admitting his role in the scam, Lee was jailed for five years and eight months.
SSG also presented another case study from July in its release yesterday.
It received feedback that a training provider offered money to applicants in exchange for course sign-ups.
The salesmen also allegedly approached the elderly for their SingPass details and got them to sign up.
SSG found that the training provider's agents had signed trainees up without their knowledge, and the trainees did not receive any training.
SSG then terminated the training provider's contract, placed it on a blacklist and recovered the funds.
SSG added that the case may be referred to the Commercial Affairs Department for further investigation.
Mr Ng Cher Pong, chief executive of SSG, said the use of technology will allow the agency to better detect fraud.
"With fraudsters devising increasingly elaborate scams, our fraud risk management system is evolving to better safeguard the funds earmarked for the skills development of our workforce," he said.
"Coupled with a firm enforcement stance, these measures will serve as a strong deterrent against potential abuse."
Mr Patrick Tay, chairman of the Government Parliamentary Committee for Manpower, said the enhancements are a positive step in preventing and detecting SkillsFuture-related fraud.
"Those who defraud and attempt to defraud should be dealt with severely as a deterrent effect to would be perpetrators," he said.
"There is a great imperative towards encouraging skills upgrading and re-skilling in light of the changing fabric of emerging jobs.
"We should not let such incidents of fraud derail us from our cause and drive towards promoting lifelong learning and individual initiated training."