Searched defs:mean (Results 76 - 100 of 182) sorted by relevance

12345678

/external/python/cpython3/Lib/
H A Dstatistics.py13 mean Arithmetic mean (average) of data.
14 harmonic_mean Harmonic mean of data.
22 Calculate the arithmetic mean ("the average") of data:
24 >>> mean([-1.0, 2.5, 3.25, 5.75])
63 If you have previously calculated the mean, you can pass it as the optional
67 >>> mu = mean(data)
82 'mean', 'mode', 'harmonic_mean',
291 def mean(data): function
292 """Return the sample arithmetic mean o
[all...]
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/
H A Dbatch_norm_op.cc69 // calculated mean and variance.
76 // space 1 & 2". They are used to pass the per-batch mean and
78 // them to the mean and variance calculated by BatchNormTraining.
86 // Directly send input to output as mean and variance in inference mode.
126 auto mean = ctx->Input(3); variable
151 b->BatchNormGrad(activations, scale, mean, var, grad_backprop,
177 // scratch2 = sum(y_backprop * (x - mean))
179 b->Mul(grad_backprop, b->Sub(activations, mean, {feature_index})),
/external/tensorflow/tensorflow/compiler/xla/service/gpu/
H A Dgpu_layout_assignment_test.cc109 // The shape of the scale, offset, mean, and variance inputs to
128 auto* mean = builder.AddInstruction( local
129 HloInstruction::CreateParameter(3, aux_shape, "mean"));
141 {operand, scale, offset, mean, variance, epsilon, feature_index},
247 // The shape of the scale, mean, and variance inputs to BatchNormGrad. These
270 auto* mean = builder.AddInstruction( local
271 HloInstruction::CreateParameter(2, scale_shape, "mean"));
286 {operand, scale, mean, var, grad_offset, epsilon,
H A Dcudnn_batchnorm_thunk.cc82 const BufferAllocation::Slice& mean,
89 mean_(mean),
166 // batch mean, and batch variance. We want to make our descriptors based on
212 const BufferAllocation::Slice& scale, const BufferAllocation::Slice& mean,
222 mean_(mean),
79 CudnnBatchNormForwardInferenceThunk( const BufferAllocation::Slice& operand, const BufferAllocation::Slice& scale, const BufferAllocation::Slice& offset, const BufferAllocation::Slice& mean, const BufferAllocation::Slice& variance, float epsilon, int64 feature_index, const BufferAllocation::Slice& output, const HloInstruction* hlo) argument
210 CudnnBatchNormBackwardThunk( const BufferAllocation::Slice& operand, const BufferAllocation::Slice& scale, const BufferAllocation::Slice& mean, const BufferAllocation::Slice& inv_stddev, const BufferAllocation::Slice& grad_output, float epsilon, int64 feature_index, const BufferAllocation::Slice& output_grad_data, const BufferAllocation::Slice& output_grad_scale, const BufferAllocation::Slice& output_grad_offset, const BufferAllocation::Slice& output_tuple, const HloInstruction* hlo) argument
/external/tensorflow/tensorflow/compiler/xla/tests/
H A Dliteral_test_util.h219 // values using a normal distribution with given mean and stddev standard
226 const Shape& shape, E* engine, T mean, T stddev);
229 // values using a normal distribution with given mean and stddev standard
236 const Shape& shape, T mean, T stddev);
363 LiteralTestUtil::CreateRandomLiteral(const Shape& shape, E* engine, T mean, argument
366 std::normal_distribution<NativeT> generator(mean, stddev);
375 LiteralTestUtil::CreateRandomLiteral(const Shape& shape, T mean, T stddev) { argument
377 return CreateRandomLiteral<type>(shape, &engine, mean, stddev);
/external/tensorflow/tensorflow/core/kernels/
H A Dbatch_norm_op.cc49 const Tensor& mean = context->input(1); variable
57 OP_REQUIRES(context, mean.dims() == 1,
58 errors::InvalidArgument("mean must be 1-dimensional",
59 mean.shape().DebugString()));
75 context->eigen_device<Device>(), input.tensor<T, 4>(), mean.vec<T>(),
99 const Tensor& mean = context->input(1); variable
107 OP_REQUIRES(context, mean.dims() == 1,
108 errors::InvalidArgument("mean must be 1-dimensional",
109 mean.shape().DebugString()));
125 {1}, 1, mean
[all...]
H A Dbatch_norm_op.h29 typename TTypes<T>::ConstVec mean,
35 const int depth = mean.dimension(0);
54 mean.reshape(one_by_depth).broadcast(rest_by_one)) *
63 mean.reshape(one_by_depth).broadcast(rest_by_one)) *
76 typename TTypes<T>::ConstVec mean,
85 const int depth = mean.dimension(0);
122 mean.reshape(one_by_depth).broadcast(rest_by_one)))
28 operator ()(const Device& d, typename TTypes<T, 4>::ConstTensor input, typename TTypes<T>::ConstVec mean, typename TTypes<T>::ConstVec var, typename TTypes<T>::ConstVec beta, typename TTypes<T>::ConstVec gamma, T variance_epsilon, bool scale_after_normalization, typename TTypes<T, 4>::Tensor output) argument
75 operator ()(const Device& d, typename TTypes<T, 4>::ConstTensor input, typename TTypes<T>::ConstVec mean, typename TTypes<T>::ConstVec var, typename TTypes<T>::ConstVec gamma, typename TTypes<T, 4>::ConstTensor out_backprop, T variance_epsilon, bool scale_after_normalization, typename TTypes<T, 4>::Tensor dx, typename TTypes<T>::Vec dm, typename TTypes<T>::Vec dv, typename TTypes<T>::Vec db, typename TTypes<T>::Vec dg, typename TTypes<T>::Vec scratch1, typename TTypes<T>::Vec scratch2) argument
H A Dadjust_contrast_op.cc220 // Calculate the mean of the inputs.
222 // Broadcast the mean into the outputs.
230 // Reduce the mean of the inputs along the image dimension, i.e. dim_1, in a
231 // 3D tensor. Effectively means(i, k) = inputs(i, :, k).mean().
233 typename TTypes<float, 2>::Tensor mean,
239 TTypes<float, 1>::Tensor mean_flat(&mean(0, 0), mean.size());
255 // Sum the input(i, :, k) into mean(i, k). Repeatedly splits the input
232 ReduceMeanAcrossImage(typename TTypes<float, 3>::ConstTensor input, typename TTypes<float, 2>::Tensor mean, typename TTypes<float, 3>::Tensor scratch) argument
H A Ddebug_ops.h290 double mean = std::numeric_limits<double>::quiet_NaN(); variable
341 mean = sum / non_inf_nan_count;
348 variance += (x - mean) * (x - mean);
367 output_tensor->vec<double>()(10) = mean;
H A Deigen_attention.h165 // Initialize the glimpse with white noise: compute the mean and sigma
173 TensorFixedSize<Scalar, Sizes<> > mean; local
174 mean.device(device) = input.template chip<3>(i)
177 .mean();
183 mean.reshape(Sizes<1, 1>()).broadcast(input_size))
185 .mean()
196 (mean.reshape(Sizes<1, 1>()).broadcast(glimpse_size) +
H A Dparameterized_truncated_normal_op_gpu.cu.cc80 T mean = means[single_mean ? 0 : batch_id]; local
87 if (Eigen::numext::isinf(minval) || maxval < mean) {
99 const T normMin = (minval - mean) / stddev;
100 const T normMax = (maxval - mean) / stddev;
150 data[offset] = z[i] * stddev + mean;
175 data[offset] = z * stddev + mean;
H A Dquantized_batch_norm_op.cc32 const float input_max, const Tensor& mean,
40 auto mean_flat = mean.flat<T1>();
46 const int depth = mean.dim_size(0);
95 const float input_max, const Tensor& mean,
103 auto mean_flat = mean.flat<T1>();
109 const int depth = mean.dim_size(0);
178 const Tensor& mean = context->input(3); variable
194 OP_REQUIRES(context, mean.dims() == 1,
195 errors::InvalidArgument("mean must be 1-dimensional",
196 mean
31 ReferenceBatchNorm(const Tensor& input, const float input_min, const float input_max, const Tensor& mean, float mean_min, float mean_max, const Tensor& var, float var_min, float var_max, const Tensor& beta, float beta_min, float beta_max, const Tensor& gamma, float gamma_min, float gamma_max, float variance_epsilon, bool scale_after_normalization, Tensor* output, float* output_min, float* output_max) argument
94 FixedPointBatchNorm(const Tensor& input, const float input_min, const float input_max, const Tensor& mean, float mean_min, float mean_max, const Tensor& var, float var_min, float var_max, const Tensor& beta, float beta_min, float beta_max, const Tensor& gamma, float gamma_min, float gamma_max, float variance_epsilon, bool scale_after_normalization, Tensor* output, float* output_min, float* output_max) argument
[all...]
/external/tensorflow/tensorflow/stream_executor/cuda/
H A Dcuda_rng.cc209 bool CUDARng::DoPopulateRandGaussianInternal(Stream *stream, ElemT mean, argument
221 func(parent_, rng_, CUDAMemoryMutable(v), element_count, mean, stddev);
232 bool CUDARng::DoPopulateRandGaussian(Stream *stream, float mean, float stddev, argument
234 return DoPopulateRandGaussianInternal(stream, mean, stddev, v,
238 bool CUDARng::DoPopulateRandGaussian(Stream *stream, double mean, double stddev, argument
240 return DoPopulateRandGaussianInternal(stream, mean, stddev, v,
/external/walt/android/WALT/app/src/main/java/org/chromium/latency/walt/
H A DUtils.java46 public static double mean(double[] x) { method in class:Utils
84 double m = mean(a);
/external/webrtc/webrtc/base/
H A Dprofiler.h97 double mean() const { return mean_; } function in class:rtc::ProfilerEvent
H A Drandom_unittest.cc60 // Expect the result to be within 3 standard deviations of the mean.
141 // Expect the result to be within 3 standard deviations of the mean,
142 // or more generally, within sigma_level standard deviations of the mean.
143 double mean = static_cast<double>(samples) / bucket_count; local
144 EXPECT_NEAR(buckets[i], mean, sigma_level * sqrt(mean));
181 // Expect the result to be within 3 standard deviations of the mean,
182 // or more generally, within sigma_level standard deviations of the mean.
183 double mean = static_cast<double>(samples) / bucket_count; local
184 EXPECT_NEAR(buckets[i], mean, sigma_leve
235 double mean = static_cast<double>(samples) / bucket_count; local
[all...]
H A Drollingaccumulator.h125 // between (0.0, 1.0], otherwise the non-weighted mean is returned.
153 double mean = sum_ * count_inv; local
154 return mean_2 - (mean * mean);
/external/guava/guava/src/com/google/common/math/
H A DLongMath.java748 * Returns the arithmetic mean of {@code x} and {@code y}, rounded toward
753 public static long mean(long x, long y) { method in class:LongMath
754 // Efficient method for computing the arithmetic mean.
/external/libchrome/base/trace_event/
H A Dmemory_dump_scheduler.cc237 uint64_t mean = 0; local
246 mean += polling_state_->last_memory_totals_kb[i];
248 mean = mean / PollingTriggerState::kMaxNumMemorySamples;
251 variance += (polling_state_->last_memory_totals_kb[i] - mean) *
252 (polling_state_->last_memory_totals_kb[i] - mean);
261 bool is_stddev_low = variance < mean / 500 * mean / 500;
265 // (mean + 3.69 * stddev) corresponds to a value that is higher than current
267 return (current_memory_total_kb - mean) * (current_memory_total_k
[all...]
/external/mesa3d/src/gallium/auxiliary/util/
H A Du_cache.c295 double mean = (double)cache->count/(double)cache->size; local
296 double stddev = sqrt(mean);
299 double z = fabs(cache->entries[i].count - mean)/stddev;
/external/opencv/cxcore/src/
H A Dcxmean.cpp165 mean[0] = scale*(double)tmp##0
171 mean[0] = t0; \
172 mean[1] = t1
179 mean[0] = t0; \
180 mean[1] = t1; \
181 mean[2] = t2
187 mean[0] = t0; \
188 mean[1] = t1; \
191 mean[2] = t0; \
192 mean[
384 CvScalar mean = {{0,0,0,0}}; local
[all...]
/external/tensorflow/tensorflow/contrib/lite/kernels/
H A Dmean.cc27 namespace mean { namespace in namespace:tflite::ops::builtin
220 } // namespace mean
223 static TfLiteRegistration r = {mean::Init, mean::Free, mean::Prepare,
224 mean::Eval<mean::kReference>};
/external/tensorflow/tensorflow/python/kernel_tests/
H A Dparameterized_truncated_normal_op_test.py40 mean = None variable in class:TruncatedNormalMoments
45 def __init__(self, mean, stddev, minval, maxval):
47 self.mean = np.double(mean)
69 dist = scipy.stats.norm(loc=self.mean, scale=self.stddev)
76 m = ((k - 1) * self.stddev**2 * m_k_minus_2 + self.mean * m_k_minus_1 -
112 def validateMoments(self, shape, mean, stddev, minval, maxval, seed=1618):
119 samples = random_ops.parameterized_truncated_normal(shape, mean, stddev,
124 expected_moments = TruncatedNormalMoments(mean, stddev, minval, maxval)
134 mean,
[all...]
/external/tensorflow/tensorflow/tools/graph_transforms/
H A Dfold_old_batch_norms.cc72 Tensor mean = GetNodeTensorAttr(mean_node, "value"); local
79 const int64 num_cols = mean.shape().dim_size(0);
102 (-mean.flat<float>()(i) * (*scale_values)[i]) + beta.flat<float>()(i);
/external/webrtc/webrtc/common_audio/
H A Daudio_converter_unittest.cc59 float mean = 0; local
65 mean += ref.channels()[i][j];
72 mean /= length;
73 variance -= mean * mean;

Completed in 3161 milliseconds

12345678