Thanks to techniques of mechanical learning, algorithms were able to search for artificial samples with gross reliability after traces of auberrish life. This means a research team about nicholas guttenberg from the tokyo institute of technology. They have used the preserved analysis method of mass spectrometry in a unusual way with a ki and were successful.
Unlike commonly ubrow, the ki did not search the measurement data for traces of the individual components, but compared the overall pattern. At 95 percent correctly they could specify whether there were traces of living organisms in the analyzed material. How that worked, the researchers do not really understand.
Do not search in detail, but the rough whole
As the team now succeeded, it wanted to devote himself with his action of that problem, which according to which we may not discover aubertirdian life, because we only look for obvious life. So we did not know if the resulting life on earth has struggled at the beginning certain directions from accident or need.
For example, the researchers relate to certain chemical processes or building blocks based on today’s life on earth. So we look for a lookout for auberdish life for about very specific molecules, even if this could be built from completely different building blocks.
When looking for such building blocks, research often sets in mass spectrometry through which the individual components of a sample can be determined. Such instruments are used in research institutes on earth, but are also active in research probes, for example on mars. The team around guttenberg now has now recorded mass spectrometric analyzes of various complex organic compounds (including in the laboratory guaranteed non-biological), of life building blocks of meteorites (which did not develop any life), from microorganisms and even unprocessed petroleum (as a uberrest long deceased living beings included). The appropriately trained ki should then put this mass spectra in "live" and "non-living" classify without looking for specific substances specifically.
The coarse part correctly stored mass spectra
At his surprise, the ki had made this task to about 95 percent correctly, writes the team. That even worked, though the raw data "significant" were simplified to simulate results less well-soluble instruments for example on a mars-rover. How exactly the classification works, must still be explored, the team also admits.
The hypothesis of the researchers is that the algorithms patterns relating to the underlying processes "recognizes". After all, musse life creates copies of yourself, non-biological processes not. Overall olfen her work "exciting perspectives" for the astrobiology, they are safe. Your scientific essay is in the trade magazine life appeared.