BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Vrije Universiteit Amsterdam//NONSGML v1.0//EN NAME:PhD defence L.P. Silvestrin METHOD:PUBLISH BEGIN:VEVENT DTSTART:20260513T134500 DTEND:20260513T151500 DTSTAMP:20260513T134500 UID:2026/phd-defence-l-p-silvestri@8F96275E-9F55-4B3F-A143-836282E12573 CREATED:20260502T103842 LOCATION:(1st floor) Auditorium, Main building De Boelelaan 1105 1081 HV Amsterdam SUMMARY:PhD defence L.P. Silvestrin X-ALT-DESC;FMTTYPE=text/html:

Efficient Machine Learn ing for Time-Varying Data

Research makes AI more reliable wit h changing data

Machine learning can handle changing and incom plete data much better than previously assumed. This is evident from research by data scientist Luis Silvestrin, who developed new methods to analyze time series of sensor data - such as measurements from ma chines or medical equipment - more reliably.

In practice, this type of data changes constantly. Sensors are adjusted, conditions flu ctuate, and important signals, such as malfunctions or medical compli cations, often occur only rarely. According to Silvestrin, standard m achine learning methods frequently fall short as a result. They impli citly assume that data remains stable, whereas in reality, this is ra rely the case.

The research shows that usable predictions do re main possible, provided that algorithms explicitly take this variabil ity into account. Silvestrin developed techniques that handle limited and evolving datasets better. These methods proved effective in indu strial applications and healthcare, among others.

Fewer malfunctions and better care decisions

The societal i mpact of these findings could be significant. In industry, companies can use the new approach to detect anomalies in machines earlier, eve n before malfunctions occur. This saves costs and prevents downtime. A concrete example is a warning system for overheating motors in conv eyor belts. The new methods help ensure that rare problems are not mi ssed, while simultaneously limiting the number of false alarms.

This also offers opportunities in hospitals. Doctors often have to m ake decisions based on limited and constantly changing patient data. The new techniques can, for example, assist in determining the right moment to remove a breathing tube in the intensive care unit, even wh en little new data is available.

AI that moves with rea lity

The core of the research is that artificial intel ligence works better when systems adapt to change, rather than assumi ng a stable world. This is relevant in a time when more and more deci sions depend on data that is incomplete, noise-sensitive, and dynamic .

Some applications of the new methods are already immediately deployable because they have been tested in realistic environments. H owever, further validation is required for broader application, depen ding on the specific sector. The research thus aligns with larger soc ietal developments, such as the rise of smart healthcare systems, a m ore reliable industry, and the growing role of AI in complex, realist ic situations.

More information on the thesis

DESCRIPTION:

Research makes AI more reliable with changing data Machine learning can handle changing and incomplete data much bett er than previously assumed. This is evident from research by data sci entist Luis Silvestrin, who developed new methods to analyze time ser ies of sensor data - such as measurements from machines or medical eq uipment - more reliably. In practice, this type of data changes const antly. Sensors are adjusted, conditions fluctuate, and important sign als, such as malfunctions or medical complications, often occur only rarely. According to Silvestrin, standard machine learning methods fr equently fall short as a result. They implicitly assume that data rem ains stable, whereas in reality, this is rarely the case. The researc h shows that usable predictions do remain possible, provided that alg orithms explicitly take this variability into account. Silvestrin dev eloped techniques that handle limited and evolving datasets better. T hese methods proved effective in industrial applications and healthca re, among others. Fewer malfunctions and better care decision s The societal impact of these findings could be significant . In industry, companies can use the new approach to detect anomalies in machines earlier, even before malfunctions occur. This saves cost s and prevents downtime. A concrete example is a warning system for o verheating motors in conveyor belts. The new methods help ensure that rare problems are not missed, while simultaneously limiting the numb er of false alarms. This also offers opportunities in hospitals. Doct ors often have to make decisions based on limited and constantly chan ging patient data. The new techniques can, for example, assist in det ermining the right moment to remove a breathing tube in the intensive care unit, even when little new data is available. AI that m oves with reality The core of the research is that artificia l intelligence works better when systems adapt to change, rather than assuming a stable world. This is relevant in a time when more and mo re decisions depend on data that is incomplete, noise-sensitive, and dynamic. Some applications of the new methods are already immediately deployable because they have been tested in realistic environments. However, further validation is required for broader application, depe nding on the specific sector. The research thus aligns with larger so cietal developments, such as the rise of smart healthcare systems, a more reliable industry, and the growing role of AI in complex, realis tic situations. More information on the thesis Efficient Mach ine Learning for Time-Varying Data END:VEVENT END:VCALENDAR