It is possible to get certain results in the estimates, but this usually involves calculating at full quantum fluid on a cluster, and if they know, the solvent choice and treatment temperature or small impurities are accurate.
However, the development of artificial intelligence has led to several impressive presentations in chemistry. It is possible to see what this is; Artificial intelligence can understand the rules of the dog, and the computer does not need the constraints given by chemistry education. A group working at Glasgow University has combined a machine-learning system powered by a robot that can conduct and analyze chemical chemistry. The result was a system that could estimate all possible treatments of the starting materials of the data.
A chemist in the bowl
Lее Cronin would send the following visual as the researcher organizing the study. Nothing we describe as a robot in this installation work is available (the researchers call it ‘bespoke’). Most of the pieces are in a burner, so that any type of compound that can escape in the system is safely removed from the environment. The containers seen on the upper right contain the starting materials and pumps, which send substances to one of the six treatment containers and can operate in parallel.
The results of these treatments can be sent for analysis. Pumps can send samples to an IR spectrometer, mass spectrometer, and a small NMR machine, this last mentioned machine is too large to fit in the burner. These provide the data that will form the “fingerprint” of a complete treatment. Comparing these with the fingerprints of the starting materials, it is possible to think that the treatment has been realized and it is possible to predict some questions about the products.
All of these may overwhelm the people of a chemist, but we do not overwhelm the person, which will judge potential treatments. Here, the machine learning algorithm comes into play. The system starts with 72 treatments with known products and guesses about the products of the treatments. From here, it selects a solution randomly based on the list of available options and finds out that they will not produce products. When the algorithm processes 10% of the total possible treatments, it can draw an idea about the state of untested treatments with more than 80%.
Apart from that, since the first treatment methods are chosen randomly, it does not affect people’s doubts as to which equipment will or will not work.
When a modеl occurs sistеm which possible tеpkimеlеr the ürünlеr the vеrеcеg the dеğеrlеndirеcеk şеkildе Setting vе is their tеstlеr the foreground .Sistеm a sеr the tеpkimеy the dеnеyеnе up study dеva I еdеbil is a sеr the tеpkimеdеn product еldе еdilmеdiğindе can stand, vеy all tеpkimеlеr can operate as dеnеyеnе.
Regardless of this grade of success, the research team has created a neural network that provides data on research literature on a class of chemistry combining two hydrocarbon chains. After working on approximately 3500 applications, it is seen that the system has passed product information about 1700 applications in the literature with only 11% error.
This system was merged with the existing test installation and operated to estimate the methods not reported in the literature. Thus, the system did not only predict which treatment will produce a product or not, it was also predicted how much product would be produced by the treatment.
All this is pretty impressive on its own. According to the authors, “we can only examine 10% of the consultants and estimate the outputs of the remaining 90% without execution” . But the system was able to catch a few surprises as well – indicating that the fingerprint of the treatment mixture is more than a simple combination of the starting materials. These treatments were further studied by human chemists, and by virtue of them ring rupture and ring formation patterns were identified.
This last issue offers a long way to go into future chemistry laboratories on how to enter such a competence. People often think that robots will come to people. In this case, however, robots seem to be taking heavy and tedious tasks in the hands of people. No sane person would think of considering the whole combination of reactives, and people would not be able to stand 24 hours a day unless they attended the dangerous café’s counter. Robots will do a good job of identifying rare situations where highly educated predictions inform us that we are getting out of the way of some of the experts.
For now, however, people know that they need to transform this information into useful chemical information, and that putting forward effective procedures can create a previously uneffective process or a product to work with more easily.