關聯預測模型建立在一個額外泛型的身份上 資料集，其中每個標識包含多個圖

關聯預測模型建立在一個額外泛型的身份上 資料集，其中每個標識包含多個圖像
變化較大內部個人。考慮兩個時 面孔下顯著不同的設置 (例如，非額葉 和額葉），我們第一次"關聯"一輸入人臉都 從設置的泛型標識日期的身份。使用關聯的 面孔，我們 generatively"預測"的外觀 一個輸入下的另一個輸入的臉上，設置工作面或 discriminatively"預測"可能性是否兩個輸入
面臨來自同一個人不是是。我們稱這兩個提議
作為"外觀-預測"的預測方法和
"可能性預測"。通過利用額外的資料 set
("memory") 和"准預測"模型中，個人才能
變化可以有效地處理。

結果 (英文) 1:

Associated forecasting model based on an additional data on the identity of the generic collection, which contains more than one image for each identity
large variation within individuals. Face when considering two dramatically different settings (for example, non-frontal and frontal lobe), the first time we "connected" type face are identified from a set of generic date status. Using the associated faces, our generatively "prediction" appearance entered under another type of face, set the face or discriminatively "prediction" possibilities are two input
faces is not from the same person. We call these two proposed
As "appearance-project" prediction method and
"probability forecasts." Through the use of extra data set
("memory") and the "prediction" model, the individual talent
change can be effectively addressed.

結果 (英文) 2:

Associated predictive model based on the identity of an additional generic data sets, each of which contains multiple images to identify
changes in the larger internal individuals. When considering the two faces under significantly different settings (for example, non-frontal and frontal), we first "association" status of an input from a set of generic human face identification date. Associated with the use of the face, we generatively "predict" the appearance of an input to another input face, set face or discriminatively "predict" whether the possibility of the two input
faces from the same person is not yes. We call these two proposals
as the "Appearance - predict" forecasting methods and
"the possibility of prediction." By using additional data set
("Memory") and "quasi-prediction" model, individuals can
effectively deal with change.

結果 (英文) 3:

The connection forecast model establishes in a status of extra pan-the material collection, each marking contains many image
change big interior individuals. When considers two under face remarkable different establishment (e.g., non-temporal lobe and temporal lobe), our first time " connection " an input face from establishment the status of pan-marking date. Face of use connection, our generatively " forecast " on outward appearance input the face under of another input, establishes the working surface or discriminatively " forecast that " possible whether two input
faced with from the same person is not. We said that these two proposed
As " outward appearance - forecast "the forecast technique and
" possibility forecasts". ("Memory") and " forecasts " in model through use extra material set
certainly, individual can the
change be able effectively to process.