Searched refs:loss (Results 1 - 3 of 3) sorted by relevance

/frameworks/ml/bordeaux/learning/multiclass_pa/native/
H A Dmulticlass_pa.cpp114 float loss = 1.0 - target_class_score + other_class_score; local
115 if (loss > 0.0) {
121 float rate = loss / twice_norm_square;
125 // VLOG(1) << "loss = " << loss << " rate = " << rate;
133 return loss;
147 float loss = 1.0 - target_class_score + other_class_score; local
148 if (loss > 0.0) {
154 float rate = loss / twice_norm_square;
158 // VLOG(1) << "loss
[all...]
/frameworks/ml/bordeaux/learning/stochastic_linear_ranker/native/
H A Dstochastic_linear_ranker.cpp37 const double loss = std::max(0.0, (1 - positive_score + negative_score)); local
45 std::min(lambda_, loss / (gradient_norm * gradient_norm));
48 loss / (gradient_norm * gradient_norm + 1 / (2 * lambda_));
101 ALOGE("Unknown rank loss type: %d", rank_loss_type_);
/frameworks/base/wifi/java/android/net/wifi/
H A DWifiWatchdogStateMachine.java51 * We now monitor the quality of the last hop on WiFi using packet loss ratio as
56 * instant packet loss, and record it as per-AP loss-to-rssi statistics. When
57 * the instant packet loss is higher than a threshold, the WiFi watchdog sends a
62 * lower packet loss, a good link detection is sent such that the WiFi
120 * Remember packet loss statistics of how many BSSIDs.
128 * Within the range, (RSSI -> packet loss %) mappings are stored.
141 * In this case, a preset/default loss value (function on RSSI) is used.
149 * Sample interval for packet loss statistics, in msec.
156 * Coefficients (alpha) for moving average for packet loss trackin
[all...]

Completed in 184 milliseconds