![]() Enter the Internet and a proliferation of cable networks, which left fewer than one out of five households watching the Tuesday night “American Idol,” the top-rated series for 2005-2006. television households tuned into “The Bill Cosby Show,” according to Nielsen Media Research. In some cases, the effect has been quite dramatic.Īs recently as 20 years ago, for example, one out of three U.S. “Sometimes that means from hits to niches, sometimes that means from new releases to back catalog. “Demand is shifting the down the tail,” Anderson said. He argues, however, that the theory applies to numerous other businesses, including appliances, foods, toys and even employment, saying offshoring is “the long tail of labor.” INTERVIEW-Questions for "Long Tail" author Chris AndersonĪnderson initially coined the phrase “long tail” to describe the new economics of media and entertainment. But Anderson notes that more people are dipping into esoteric fare as the Internet and other technology make it easier to find, produce and distribute. The title refers to “long-tailed distribution,” a statistical term describing how a typical demand curve starts with a large “head” - the hits - and trails off into a long “tail” - everything else.Ĭonventional wisdom says that the top 20 percent of a company’s products account for 80 percent of sales. While a cooking program or a nature documentary is unlikely to pose a challenge to “Desperate Housewives,” in the aggregate, niche markets are becoming a force to be reckoned with in a number of industries.Ĭhris Anderson, the editor in chief of Wired Magazine, made these points in an October 2004 article for his magazine and expanded them into a book, “The Long Tail,” published in July. TV show “Desperate Housewives” may be dying to discuss last night’s episode, but the odds are that their friends and co-workers watched something else.Īs technology makes production, storage and distribution of entertainment easier, choices are abounding, and so are audiences for alternatives to hit TV shows - and songs, books or movies. The consistent state-of-the-art experiments on the long-tailed CIFAR-100, ImageNet, Places and iNaturalist demonstrate the benefit of the discovered super-class graph for dealing with long-tailed distributions.NEW YORK, Dec 11 (Reuters) - Fans of the popular U.S. By doing so, we obtain a more robust super-class graph that further improves the long-tailed recognition performance. Moreover, we propose to meta-learn the super-class graph under the supervision of a prototype graph constructed from a small amount of imbalanced data. Through message passing on the super-class graph, image representations are rectified and refined by attending to the most relevant entities based on the semantic similarity among their super-classes. We learn to construct the super-class graph to guide the representation learning to deal with long-tailed distributions. This phenomenon motivated us to propose SuperDisco, an algorithm that discovers super-class representations for long-tailed recognition using a graph model. Humans, by contrast, effortlessly handle the long-tailed recognition challenge, since they can learn the tail representation based on different levels of semantic abstraction, making the learned tail features more discriminative. Modern image classifiers perform well on populated classes, while degrading considerably on tail classes with only a few instances. The CVPR 2023 paper SuperDisco: Super-Class Discovery Improves Visual Recognition for the Long-Tail by Yingjun Du, Jiayi Shen, Xiantong Zhen, and Cees Snoek is now available.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |