On a recent Winter weekend, I was snowed-in in my house. To use some time productively, I dusted off some old JavaScript code I wrote. The code implements a complete neural network classifier from scratch. The goal of the demo multi-class classifier NN is to predict the size of the company (small, medium, large) that a student works for after graduation, based on their sex (M, F), science units taken, major (finance, geology, history), and parents income.
The data is synthetic. It looks like:
-1 0.39 0 0 1 0.3570 0 1 0 1 0.51 1 0 0 0.8610 0 1 0 -1 0.19 0 0 1 0.6550 1 0 0 1 0.24 0 1 0 0.5700 0 0 1 1 0.22 1 0 0 0.8970 0 1 0 . . .
The first column is sex (M = -1, F = 1), the next column is number of science units taken, normalized by dividing by 100. The next three columns are the college major, one-hot encoded (finance = 100, geology = 010, history = 001), the sixth column is parent’s income, normalized by dividing by 100,000. The last three columns are the company size to predict (small = 100, medium = 010, large = 001).
The code is complex and involves every aspect of neural networks: network architecture, input-output, activation functions, mean squared error/loss, gradient descent training, softmax, classification accuracy, saving model weights and biases, and many other sub-topics. But the point is that neural networks aren’t magic — they’re just ideas and code, even if the ideas and code are complicated.
My demo runs in a shell and is based on the node.js framework. The demo code could be adapted to run in a browser, by replacing console.log() statements with some sort of display on a Web page in an HTML element.
A mildly annoying aspect of node.js file IO is that there is no easy way to do syncronous text file read. At first I though I must be completely missing something, but in node.js you just can’t read a file line-by-line without working at a very low level and buffering characters or bytes. Ugh.
Here are a few representative code snippets. The complete demo code is listed below.
. . .
class NeuralNet
{
constructor(numInput, numHidden, numOutput, seed)
{
this.rnd = new U.Erratic(seed);
this.ni = numInput;
. . .
this.iNodes = U.vecMake(this.ni, 0.0);
this.hNodes = U.vecMake(this.nh, 0.0);
. . .
this.initWeights();
}
. . .
}
. . .
// compute output node signals
for (let k = 0; k < this.no; ++k) { // softmax
let derivative = (1 - this.oNodes[k]) * this.oNodes[k];
// E=(t-o)^2
oSignals[k] = derivative * (this.oNodes[k] - Y[k]);
}
. . .
// compute hidden-to-output weight gradients using output signals
for (let j = 0; j < this.nh; ++j) {
for (let k = 0; k < this.no; ++k) {
hoGrads[j][k] = oSignals[k] * this.hNodes[j];
}
}
. . .
Good fun!
There are several roller coaster model kits you can buy (CoasterDynamix, Lego, Thames and Kosmos, etc.) But Adam Throgmorton has created beautiful working model roller coasters from scratch — more interesting and beautiful than a kit model.
Demo code below. Very long. Replace “lt” (less-than), “gt”, “lte”, “gte” with symbols — my blog editor chokes on those symbols.
// students_nn.js
// ES6, node.js
// predict company size (S,M,L) from sex, science units, major, parents income.
let U = require("../Utilities/utilities_lib.js");
let FS = require("fs");
// =============================================================================
class NeuralNet
{
constructor(numInput, numHidden, numOutput, seed)
{
this.rnd = new U.Erratic(seed);
this.ni = numInput;
this.nh = numHidden;
this.no = numOutput;
this.iNodes = U.vecMake(this.ni, 0.0);
this.hNodes = U.vecMake(this.nh, 0.0);
this.oNodes = U.vecMake(this.no, 0.0);
this.ihWeights = U.matMake(this.ni, this.nh, 0.0);
this.hoWeights = U.matMake(this.nh, this.no, 0.0);
this.hBiases = U.vecMake(this.nh, 0.0);
this.oBiases = U.vecMake(this.no, 0.0);
this.initWeights();
}
initWeights()
{
let lo = -0.01;
let hi = 0.01;
for (let i = 0; i "lt" this.ni; ++i) {
for (let j = 0; j "lt" this.nh; ++j) {
this.ihWeights[i][j] = (hi - lo) * this.rnd.next() + lo;
}
}
for (let j = 0; j "lt" this.nh; ++j) {
for (let k = 0; k "lt" this.no; ++k) {
this.hoWeights[j][k] = (hi - lo) * this.rnd.next() + lo;
}
}
}
eval(X)
{
let hSums = U.vecMake(this.nh, 0.0);
let oSums = U.vecMake(this.no, 0.0);
this.iNodes = X;
for (let j = 0; j "lt" this.nh; ++j) {
for (let i = 0; i "lt" this.ni; ++i) {
hSums[j] += this.iNodes[i] * this.ihWeights[i][j];
}
hSums[j] += this.hBiases[j];
this.hNodes[j] = U.hyperTan(hSums[j]);
}
for (let k = 0; k "lt" this.no; ++k) {
for (let j = 0; j "lt" this.nh; ++j) {
oSums[k] += this.hNodes[j] * this.hoWeights[j][k];
}
oSums[k] += this.oBiases[k];
}
this.oNodes = U.softmax(oSums);
let result = [];
for (let k = 0; k "lt" this.no; ++k) {
result[k] = this.oNodes[k];
}
return result;
} // eval()
setWeights(wts)
{
// order: ihWts, hBiases, hoWts, oBiases
let p = 0;
for (let i = 0; i "lt" this.ni; ++i) {
for (let j = 0; j "lt" this.nh; ++j) {
this.ihWeights[i][j] = wts[p++];
}
}
for (let j = 0; j "lt" this.nh; ++j) {
this.hBiases[j] = wts[p++];
}
for (let j = 0; j "lt" this.nh; ++j) {
for (let k = 0; k "lt" this.no; ++k) {
this.hoWeights[j][k] = wts[p++];
}
}
for (let k = 0; k "lt" this.no; ++k) {
this.oBiases[k] = wts[p++];
}
} // setWeights()
getWeights()
{
// order: ihWts, hBiases, hoWts, oBiases
let numWts = (this.ni * this.nh) + this.nh +
(this.nh * this.no) + this.no;
let result = U.vecMake(numWts, 0.0);
let p = 0;
for (let i = 0; i "lt" this.ni; ++i) {
for (let j = 0; j "lt" this.nh; ++j) {
result[p++] = this.ihWeights[i][j];
}
}
for (let j = 0; j "lt" this.nh; ++j) {
result[p++] = this.hBiases[j];
}
for (let j = 0; j "lt" this.nh; ++j) {
for (let k = 0; k "lt" this.no; ++k) {
result[p++] = this.hoWeights[j][k];
}
}
for (let k = 0; k "lt" this.no; ++k) {
result[p++] = this.oBiases[k];
}
return result;
} // getWeights()
shuffle(v)
{
// Fisher-Yates
let n = v.length;
for (let i = 0; i "lt" n; ++i) {
let r = this.rnd.nextInt(i, n);
let tmp = v[r];
v[r] = v[i];
v[i] = tmp;
}
}
train(trainX, trainY, lrnRate, maxEpochs)
{
let hoGrads = U.matMake(this.nh, this.no, 0.0);
let obGrads = U.vecMake(this.no, 0.0);
let ihGrads = U.matMake(this.ni, this.nh, 0.0);
let hbGrads = U.vecMake(this.nh, 0.0);
let oSignals = U.vecMake(this.no, 0.0);
let hSignals = U.vecMake(this.nh, 0.0);
let n = trainX.length; // 160
let indices = U.arange(n); // [0,1,..,159]
let freq = Math.trunc(maxEpochs / 10);
for (let epoch = 0; epoch "lt" maxEpochs; ++epoch) {
this.shuffle(indices); //
for (let ii = 0; ii "lt" n; ++ii) { // each item
let idx = indices[ii];
let X = trainX[idx];
let Y = trainY[idx];
this.eval(X); // output stored in this.oNodes
// compute output node signals
for (let k = 0; k "lt" this.no; ++k) {
let derivative = (1 - this.oNodes[k]) * this.oNodes[k]; // softmax
oSignals[k] = derivative * (this.oNodes[k] - Y[k]); // E=(t-o)^2
}
// compute hidden-to-output weight gradients using output signals
for (let j = 0; j "lt" this.nh; ++j) {
for (let k = 0; k "lt" this.no; ++k) {
hoGrads[j][k] = oSignals[k] * this.hNodes[j];
}
}
// compute output node bias gradients using output signals
for (let k = 0; k "lt" this.no; ++k) {
obGrads[k] = oSignals[k] * 1.0; // 1.0 dummy input can be dropped
}
// compute hidden node signals
for (let j = 0; j "lt" this.nh; ++j) {
let sum = 0.0;
for (let k = 0; k "lt" this.no; ++k) {
sum += oSignals[k] * this.hoWeights[j][k];
}
let derivative = (1 - this.hNodes[j]) * (1 + this.hNodes[j]); // tanh
hSignals[j] = derivative * sum;
}
// compute input-to-hidden weight gradients using hidden signals
for (let i = 0; i "lt" this.ni; ++i) {
for (let j = 0; j "lt" this.nh; ++j) {
ihGrads[i][j] = hSignals[j] * this.iNodes[i];
}
}
// compute hidden node bias gradients using hidden signals
for (let j = 0; j "lt" this.nh; ++j) {
hbGrads[j] = hSignals[j] * 1.0; // 1.0 dummy input can be dropped
}
// update input-to-hidden weights
for (let i = 0; i "lt" this.ni; ++i) {
for (let j = 0; j "lt" this.nh; ++j) {
let delta = -1.0 * lrnRate * ihGrads[i][j];
this.ihWeights[i][j] += delta;
}
}
// update hidden node biases
for (let j = 0; j "lt" this.nh; ++j) {
let delta = -1.0 * lrnRate * hbGrads[j];
this.hBiases[j] += delta;
}
// update hidden-to-output weights
for (let j = 0; j "lt" this.nh; ++j) {
for (let k = 0; k "lt" this.no; ++k) {
let delta = -1.0 * lrnRate * hoGrads[j][k];
this.hoWeights[j][k] += delta;
}
}
// update output node biases
for (let k = 0; k "lt" this.no; ++k) {
let delta = -1.0 * lrnRate * obGrads[k];
this.oBiases[k] += delta;
}
} // ii
if (epoch % freq == 0) {
let mse = this.meanSqErr(trainX, trainY).toFixed(4);
let acc = this.accuracy(trainX, trainY).toFixed(4);
let s1 = "epoch: " + epoch.toString().padStart(6, ' ');
let s2 = " MSE = " + mse.toString().padStart(8, ' ');
let s3 = " acc = " + acc.toString();
console.log(s1 + s2 + s3);
}
} // epoch
} // train()
meanCrossEntErr(dataX, dataY)
{
let sumCEE = 0.0; // sum of the cross entropy errors
for (let i = 0; i "lt" dataX.length; ++i) { // each data item
let X = dataX[i];
let Y = dataY[i]; // target output like (0, 1, 0)
let oupt = this.eval(X); // computed like (0.23, 0.66, 0.11)
let idx = U.argmax(Y); // find location of the 1 in target
sumCEE += Math.log(oupt[idx]);
}
sumCEE *= -1;
return sumCEE / dataX.length;
}
meanSqErr(dataX, dataY)
{
let sumSE = 0.0;
for (let i = 0; i "lt" dataX.length; ++i) { // each data item
let X = dataX[i];
let Y = dataY[i]; // target output like (0, 1, 0)
let oupt = this.eval(X); // computed like (0.23, 0.66, 0.11)
for (let k = 0; k "lt" this.no; ++k) {
let err = Y[k] - oupt[k] // target - computed
sumSE += err * err;
}
}
return sumSE / dataX.length;
}
accuracy(dataX, dataY)
{
let nc = 0; let nw = 0;
for (let i = 0; i "lt" dataX.length; ++i) { // each data item
let X = dataX[i];
let Y = dataY[i]; // target output like (0, 1, 0)
let oupt = this.eval(X); // computed like (0.23, 0.66, 0.11)
let computedIdx = U.argmax(oupt);
let targetIdx = U.argmax(Y);
if (computedIdx == targetIdx) {
++nc;
}
else {
++nw;
}
}
return nc / (nc + nw);
}
saveWeights(fn)
{
let wts = this.getWeights();
let n = wts.length;
let s = "";
for (let i = 0; i "lt" n-1; ++i) {
s += wts[i].toString() + ",";
}
s += wts[n-1];
FS.writeFileSync(fn, s);
}
loadWeights(fn)
{
let n = (this.ni * this.nh) + this.nh + (this.nh * this.no) + this.no;
let wts = U.vecMake(n, 0.0);
let all = FS.readFileSync(fn, "utf8");
let strVals = all.split(",");
let nn = strVals.length;
if (n != nn) {
throw("Size error in NeuralNet.loadWeights()");
}
for (let i = 0; i "lt" n; ++i) {
wts[i] = parseFloat(strVals[i]);
}
this.setWeights(wts);
}
} // NeuralNet
// =============================================================================
function main()
{
process.stdout.write("\033[0m"); // reset
process.stdout.write("\x1b[1m" + "\x1b[37m"); // bright white
console.log("\nBegin Students Data demo with JavaScript ");
console.log("Goal is to predict size of company after graduation ");
// 1. load data
// raw data looks like: male 29 finance 38400.00 medium
// norm data looks like: -1 0.29 1 0 0 0.3840 0 1 0
let trainX = U.loadTxt(".\\Data\\students_train.txt", "\t", [0,1,2,3,4,5]);
let trainY = U.loadTxt(".\\Data\\students_train.txt", "\t", [6,7,8]);
let testX = U.loadTxt(".\\Data\\students_test.txt", "\t", [0,1,2,3,4,5]);
let testY = U.loadTxt(".\\Data\\students_test.txt", "\t", [6,7,8]);
// 2. create network
console.log("\nCreating a 6-40-3 tanh, softmax NN for Students dataset ");
let seed = 1;
let nn = new NeuralNet(6, 40, 3, seed);
// 3. train network
let lrnRate = 0.005;
let maxEpochs = 4000;
console.log("\nStarting training with learning rate = 0.005 ");
nn.train(trainX, trainY, lrnRate, maxEpochs);
console.log("Training complete ");
// 4. evaluate model
let trainAcc = nn.accuracy(trainX, trainY);
let testAcc = nn.accuracy(testX, testY);
console.log("\nAccuracy on training data = " +
trainAcc.toFixed(4).toString());
console.log("Accuracy on test data = " +
testAcc.toFixed(4).toString());
// 5. save trained model
fn = ".\\Models\\students_wts.txt";
console.log("\nSaving model weights and biases to: ");
console.log(fn);
nn.saveWeights(fn);
// 6. use trained model
let unknownRaw = ["male", 46, "finance", 66400.00];
// sex, science units, major, parents income
let unknownNorm = [-1, 0.46, 0, 0, 1, 0.6640];
let predicted = nn.eval(unknownNorm);
console.log("\nRaw features of student to predict: ");
console.log(unknownRaw);
console.log("\nNormalized features of student to predict: ");
U.vecShow(unknownNorm, 4, 12);
console.log("\nPredicted pseudo-probabilities: ");
U.vecShow(predicted, 4, 12);
let companySizes = ["small", "medium", "large"];
let predIdx = U.argmax(predicted);
let predSize = companySizes[predIdx];
console.log("Predicted company size = " + predSize);
process.stdout.write("\033[0m"); // reset
console.log("\nEnd demo");
} // main()
main();
File students_train.txt. Original is tab-delimited, this has three spaces.
-1 0.39 0 0 1 0.3570 0 1 0 1 0.51 1 0 0 0.8610 0 1 0 1 0.52 0 0 1 0.6530 0 1 0 -1 0.36 0 1 0 0.7300 0 1 0 1 0.24 0 1 0 0.5700 0 0 1 1 0.27 1 0 0 0.7610 0 0 1 -1 0.19 0 0 1 0.6550 1 0 0 1 0.22 1 0 0 0.8970 0 1 0 -1 0.39 0 0 1 0.8230 0 0 1 1 0.63 1 0 0 0.2120 1 0 0 1 0.34 0 1 0 0.3780 0 1 0 -1 0.22 0 1 0 0.8500 1 0 0 1 0.35 0 0 1 0.8660 0 0 1 -1 0.33 1 0 0 0.8330 0 1 0 1 0.45 1 0 0 0.7680 0 1 0 1 0.42 1 0 0 0.4130 0 1 0 -1 0.33 1 0 0 0.2980 0 1 0 1 0.25 0 0 1 0.6010 0 1 0 -1 0.31 1 0 0 0.4370 1 0 0 1 0.27 0 1 0 0.5750 0 0 1 1 0.48 0 1 0 0.7570 0 1 0 -1 0.64 1 0 0 0.5740 0 0 1 1 0.61 1 0 0 0.2970 1 0 0 1 0.54 0 0 1 0.4170 1 0 0 1 0.29 0 1 0 0.8570 1 0 0 1 0.51 0 0 1 0.8000 0 1 0 1 0.55 0 0 1 0.2160 1 0 0 1 0.42 0 1 0 0.5610 1 0 0 1 0.22 0 1 0 0.2760 0 0 1 1 0.68 1 0 0 0.6810 1 0 0 -1 0.61 0 1 0 0.3940 0 0 1 -1 0.34 0 0 1 0.2550 0 1 0 -1 0.25 0 0 1 0.6230 1 0 0 -1 0.31 1 0 0 0.4300 0 1 0 1 0.43 0 0 1 0.8510 0 1 0 1 0.58 1 0 0 0.8850 0 0 1 -1 0.55 1 0 0 0.7180 0 0 1 -1 0.43 1 0 0 0.7370 0 1 0 -1 0.43 0 0 1 0.8130 0 1 0 -1 0.21 0 1 0 0.7700 1 0 0 1 0.55 0 0 1 0.7330 1 0 0 1 0.64 1 0 0 0.5460 1 0 0 -1 0.41 0 1 0 0.5250 0 1 0 1 0.64 0 0 1 0.4540 1 0 0 -1 0.56 0 0 1 0.4040 0 0 1 1 0.31 0 0 1 0.2050 0 1 0 -1 0.65 0 0 1 0.6860 0 0 1 1 0.55 0 0 1 0.3180 1 0 0 -1 0.25 0 1 0 0.7050 1 0 0 1 0.46 0 0 1 0.8180 0 1 0 -1 0.36 0 1 0 0.5260 1 0 0 1 0.52 1 0 0 0.5820 0 1 0 1 0.61 0 0 1 0.6940 1 0 0 1 0.57 0 0 1 0.5740 1 0 0 -1 0.46 1 0 0 0.2240 0 1 0 -1 0.62 0 1 0 0.2460 0 0 1 1 0.55 0 0 1 0.8530 1 0 0 -1 0.22 0 0 1 0.6800 0 1 0 -1 0.51 0 1 0 0.3750 1 0 0 -1 0.32 1 0 0 0.4800 0 1 0 -1 0.21 0 0 1 0.6710 1 0 0 1 0.44 1 0 0 0.6000 0 1 0 1 0.46 1 0 0 0.5290 0 1 0 1 0.62 1 0 0 0.2320 1 0 0 1 0.57 1 0 0 0.6740 1 0 0 -1 0.67 0 0 1 0.3930 0 0 1 1 0.29 0 1 0 0.4860 0 0 1 1 0.53 0 1 0 0.6190 1 0 0 -1 0.44 0 1 0 0.8680 0 1 0 1 0.46 1 0 0 0.2300 0 1 0 -1 0.21 1 0 0 0.3250 0 1 0 -1 0.38 0 1 0 0.7450 0 1 0 1 0.52 1 0 0 0.5980 0 1 0 1 0.33 1 0 0 0.7070 0 1 0 -1 0.33 1 0 0 0.7810 0 1 0 1 0.26 1 0 0 0.3690 1 0 0 1 0.58 0 1 0 0.2750 1 0 0 1 0.43 0 0 1 0.8880 0 1 0 -1 0.46 0 1 0 0.8900 1 0 0 1 0.61 0 1 0 0.5990 1 0 0 -1 0.42 0 1 0 0.6630 0 1 0 -1 0.56 0 0 1 0.2700 0 0 1 -1 0.62 1 0 0 0.3720 0 0 1 -1 0.51 0 1 0 0.2990 0 1 0 1 0.47 0 0 1 0.8090 0 1 0 -1 0.67 1 0 0 0.3080 0 0 1 -1 0.41 0 0 1 0.6850 0 1 0 1 0.42 1 0 0 0.5000 0 1 0 1 0.64 0 1 0 0.8040 1 0 0 -1 0.47 0 1 0 0.2580 0 0 1 1 0.45 1 0 0 0.8300 0 1 0 -1 0.25 0 0 1 0.6980 1 0 0 1 0.38 0 1 0 0.8280 1 0 0 1 0.55 0 0 1 0.2040 0 1 0 -1 0.44 0 1 0 0.7880 0 1 0 1 0.33 0 1 0 0.6400 0 1 0 1 0.34 0 0 1 0.6130 0 1 0 1 0.27 1 0 0 0.7360 0 0 1 1 0.32 1 0 0 0.2340 0 1 0 1 0.42 0 0 1 0.5350 0 1 0 -1 0.24 0 0 1 0.5490 1 0 0 1 0.42 1 0 0 0.8570 0 1 0 1 0.25 0 0 1 0.2920 0 0 1 1 0.51 1 0 0 0.8000 0 1 0 -1 0.55 1 0 0 0.6180 0 0 1 1 0.44 0 1 0 0.4510 0 0 1 -1 0.18 0 1 0 0.4430 1 0 0 -1 0.67 1 0 0 0.4860 0 0 1 1 0.45 0 0 1 0.8700 0 1 0 1 0.48 0 1 0 0.4830 0 1 0 -1 0.25 1 0 0 0.2170 0 1 0 -1 0.67 0 1 0 0.7670 0 1 0 1 0.37 0 0 1 0.2670 0 1 0 -1 0.32 0 1 0 0.7700 0 1 0 1 0.48 0 1 0 0.2260 0 1 0 -1 0.66 0 0 1 0.2990 0 0 1 1 0.61 0 1 0 0.3030 1 0 0 -1 0.58 0 0 1 0.6770 0 1 0 1 0.19 0 1 0 0.3010 0 0 1 1 0.38 0 0 1 0.6840 0 1 0 -1 0.27 0 1 0 0.6120 0 1 0 1 0.42 0 1 0 0.5140 0 1 0 1 0.61 0 1 0 0.7200 1 0 0 -1 0.27 0 0 1 0.8620 1 0 0 1 0.29 1 0 0 0.6460 1 0 0 -1 0.43 0 1 0 0.5990 0 1 0 1 0.48 0 1 0 0.8250 0 1 0 1 0.27 0 0 1 0.6140 0 0 1 -1 0.44 0 1 0 0.8910 1 0 0 1 0.23 1 0 0 0.4550 0 0 1 -1 0.36 1 0 0 0.8190 0 0 1 1 0.64 0 0 1 0.3130 1 0 0 1 0.29 0 0 1 0.7660 0 0 1 -1 0.33 0 1 0 0.2010 0 1 0 -1 0.66 1 0 0 0.2350 0 0 1 -1 0.21 0 0 1 0.6740 1 0 0 1 0.27 0 1 0 0.6540 0 0 1 1 0.29 0 1 0 0.6900 0 0 1 -1 0.31 0 1 0 0.8330 0 1 0 1 0.36 0 0 1 0.8130 0 1 0 1 0.49 1 0 0 0.8170 0 1 0 -1 0.28 0 1 0 0.2110 1 0 0 -1 0.43 0 0 1 0.7810 0 1 0 -1 0.46 1 0 0 0.7460 0 1 0 1 0.57 0 1 0 0.2330 1 0 0 -1 0.52 0 0 1 0.8910 0 1 0 -1 0.31 0 0 1 0.8700 0 1 0 -1 0.55 0 1 0 0.5860 0 0 1 1 0.51 0 1 0 0.5340 0 1 0 1 0.48 1 0 0 0.4510 0 1 0 -1 0.22 0 0 1 0.2970 1 0 0 1 0.59 0 0 1 0.5920 1 0 0 1 0.34 0 1 0 0.8240 0 0 1 -1 0.64 0 1 0 0.6300 0 0 1 1 0.29 0 0 1 0.8700 0 0 1 -1 0.34 1 0 0 0.7480 0 1 0 -1 0.61 0 1 0 0.2100 0 0 1 1 0.64 0 0 1 0.4450 1 0 0 -1 0.29 0 1 0 0.6360 1 0 0 1 0.63 1 0 0 0.4700 1 0 0
File students_test.txt
-1 0.29 1 0 0 0.6690 1 0 0 -1 0.51 0 1 0 0.4330 0 1 0 -1 0.24 0 0 1 0.8230 1 0 0 1 0.48 1 0 0 0.2080 0 1 0 1 0.18 0 1 0 0.3860 1 0 0 1 0.18 0 1 0 0.5110 0 0 1 1 0.33 1 0 0 0.3470 0 0 1 -1 0.22 0 0 1 0.2600 1 0 0 1 0.29 0 0 1 0.4920 0 0 1 -1 0.44 0 0 1 0.8490 1 0 0 -1 0.65 0 0 1 0.8540 1 0 0 -1 0.56 0 1 0 0.8940 0 0 1 -1 0.52 0 0 1 0.4890 0 1 0 -1 0.29 1 0 0 0.4560 1 0 0 -1 0.47 1 0 0 0.7480 0 1 0 1 0.68 0 1 0 0.6010 0 0 1 1 0.31 0 0 1 0.2910 0 1 0 1 0.61 1 0 0 0.8320 0 0 1 1 0.19 1 0 0 0.4750 0 0 1 1 0.38 0 0 1 0.8060 0 1 0 -1 0.26 0 1 0 0.7980 1 0 0 1 0.61 1 0 0 0.7250 1 0 0 1 0.41 0 1 0 0.2430 0 1 0 -1 0.49 0 1 0 0.3900 0 1 0 1 0.56 0 1 0 0.2680 1 0 0 -1 0.48 1 0 0 0.8220 0 1 0 1 0.52 0 1 0 0.5930 0 0 1 -1 0.18 0 1 0 0.4260 1 0 0 -1 0.56 0 0 1 0.6380 0 0 1 -1 0.52 1 0 0 0.8210 0 1 0 -1 0.18 1 0 0 0.2110 0 1 0 -1 0.58 0 1 0 0.2430 0 0 1 -1 0.39 1 0 0 0.7430 0 1 0 -1 0.46 0 1 0 0.5290 0 1 0 -1 0.41 1 0 0 0.2320 0 1 0 -1 0.61 0 1 0 0.2080 0 0 1 1 0.36 1 0 0 0.5360 0 0 1 1 0.44 0 1 0 0.7420 0 1 0 1 0.28 0 1 0 0.8020 0 0 1 1 0.54 0 0 1 0.4730 1 0 0
.NET Test Automation Recipes
Software Testing
SciPy Programming Succinctly
Keras Succinctly
R Programming
2026 Visual Studio Live
2025 Summer MLADS Conference
2026 DevIntersection Conference
2025 Machine Learning Week
2025 Ai4 Conference
2026 G2E Conference
2026 iSC West Conference
You must be logged in to post a comment.