September 7，"2019 Machine Smart Forum" is held at Boston University, USA。Scholars from many universities in the United States studying artificial intelligence -related majors explore the problems and potential of machines deep learning ability and potential。

Scholar Vasanth Sarathy, a scholar at Tavz University, said，A major challenge of the deep learning of machine is the simulation and understanding of human language。Take the pronouns Bet365 app download as an example，Pronoun，But machine learning requires a lot of common sense knowledge accumulation and corresponding reasoning，Even the most advanced statistical methods and the reasoning system based on big data knowledge，Not enough to make the machine fully handle these tasks，As a result, the machine correctly understands that human speaking in learning still has many difficulties。For this，Saras and their teams through a method called "Uncertainty Modeling"，Try to let machine deep learning human language，Gradually eliminate the viaing of pronouns，and develop a multi -reasoning architecture，Improve the accuracy of machines for human language learning。

Bet365 lotto review The memory of images and videos of the machine has become a recent research hotspot。Anelise Newman, a cognitive scientist at Massachusetts Bet365 lotto review Institute of Technology, said，Human visual memory is a very complicated process: on the one hand，Human vision has the ability to remember a lot of images; on the other hand，People will remember or forget similar or similar image features，This relatively complex human visual memory encoding and discarding process，Also bring challenges to machine learning。

Newman emphasized，Studies on visual memory are limited due to lack of multiple benchmark data，Slow progress。To break the problem of the visual memory of the machine，Massachusetts Bet365 lotto review Institute of Technology launched a new experimental program，to solve the mystery of human memory of human beings，and for this, an annotated image memory data set。After fine -tuning the relevant depth features，Researchers have designed a memory level that is basically close to human machine image memory model，Make the machine have deep learning ability in the field of visual memory。

Newman explanation name，Research indicates，It is possible to predict that human cognitive capabilities in the field of computer vision are possible。Specifically，In related applications，Large -scale understanding and memory video and images can help people make more attractive videos，and bet365 live casino games predict the popularity of the video。Scholars of the participating scholars，In the future, video production may no longer rely only on the skills and aesthetic level of individual creators，The deep learning and prediction ability of the machine in this area，It will make the video of the machine make more in line with the needs of the audience，And "successful sales" is more predictable。

(Our newspaper Boston, September 15th)