Machine learning algorithms can be trained to distinguish wolves from dogs in photographs, but sometimes these algorithms learn incorrectly by assuming that any animal pictured with snow in the background is a wolf (bottom right). Two School scientists argue that scientific applications of machine learning also need to be monitored to ensure they are working properly.
Michael Keiser, PhD, and Kangway Chuang, PhD, want to use machine learning to speed the pace of drug discovery. By digging into the work of another lab, the pair realized how machine learning could lead scientists astray—and came up with methods to avoid its worst pitfalls.
A computer model can now give scientists clues about how different chemicals could be attracted to each other and form more complicated molecules. This information, in turn, could shed light on how the universe was formed and how life emerged.
Michael Marletta, PhD, who completed his PhD in pharmaceutical chemistry at UCSF in 1977, was elected to the National Academy of Sciences on April 25, 2006 at the Academy's 143rd annual meeting in Washington,
For the 2nd consecutive year, UCSF topped all United States universities and colleges in both total and federally financed spending for chemistry research and development. These results reflect data for 2003 from the National Science Foundation. These are the most recent statistics available as of...