public int PredictionEnTransformXor(Foreseer foreseer)
{
IntField orig = this.Clone();
foreseer.Initialize(orig);
int wrongPredictions = 0;
int p = 0;
for (int y = 0; y < Height; y++)
{
for (int x = 0; x < Width; x++, p++)
{
int predicted = foreseer.Foresee(orig, x, y);
Data[p] = orig.Data[p] ^ predicted;
if (Data[p] != 0)
{
wrongPredictions++;
}
foreseer.Learn(orig, x, y, orig.Data[p], predicted);
}
}
return(wrongPredictions);
}
public override int Foresee(IntField image, int x, int y)
{
_curArea = ExtractArea(image, x, y);
if (_curArea == null)
{
return(_fallback.Foresee(image, x, y));
}
if (_history.ContainsKey(_curArea))
{
return(_history[_curArea].Predicted());
}
else
{
return(_fallback.Foresee(image, x, y));
}
}
public override int Foresee(IntField image, int x, int y)
{
_curArea = ExtractArea(image, x, y, out _curMostFreqSymbol, out _curSecondFreqSymbol);
if (_curArea == null)
{
return(_fallback.Foresee(image, x, y));
}
if (_history.ContainsKey(_curArea))
{
return(_history[_curArea].MostFrequent > _history[_curArea].Other ? _curMostFreqSymbol : _curSecondFreqSymbol);
}
else
{
return(_fallback.Foresee(image, x, y));
}
}
public void PredictionDeTransformXor(Foreseer foreseer)
{
IntField diff = this.Clone();
foreseer.Initialize(this);
int p = 0;
for (int y = 0; y < Height; y++)
{
for (int x = 0; x < Width; x++, p++)
{
int predicted = foreseer.Foresee(this, x, y);
Data[p] = diff.Data[p] ^ predicted;
foreseer.Learn(this, x, y, this.Data[p], predicted);
}
}
}
public void Prediction(Foreseer foreseer)
{
IntField orig = this.Clone();
foreseer.Initialize(orig);
int p = 0;
for (int y = 0; y < Height; y++)
{
for (int x = 0; x < Width; x++, p++)
{
int predicted = foreseer.Foresee(orig, x, y);
Data[p] = predicted;
foreseer.Learn(orig, x, y, orig.Data[p], predicted);
}
}
}
public override int Foresee(IntField image, int x, int y)
{
double conf0 = 0, conf1 = 0, conf2 = 0, conf3 = 0;
foreach (var size in _sizes)
{
ulong hash = HashImageRegion(image, x, y, size.Width, size.Height);
uint c0, c1, c2, c3;
_hashtable.GetCounts(hash, out c0, out c1, out c2, out c3);
double total = c0 + c1 + c2 + c3;
if (total == 0)
{
continue;
}
double wt = Math.Sqrt(size.Width * size.Height);
conf0 += c0 / total * wt;
conf1 += c1 / total * wt;
conf2 += c2 / total * wt;
conf3 += c3 / total * wt;
}
if (conf0 == 0 && conf1 == 0 && conf2 == 0 && conf3 == 0)
{
return(_fallback.Foresee(image, x, y));
}
else if (conf0 >= conf1 && conf0 >= conf2 && conf0 >= conf3)
{
return(0);
}
else if (conf1 >= conf2 && conf1 >= conf3)
{
return(1);
}
else if (conf2 >= conf3)
{
return(2);
}
else
{
return(3);
}
}
public void PredictionEnTransformDiff(Foreseer foreseer, int modulus)
{
IntField orig = this.Clone();
foreseer.Initialize(orig);
int p = 0;
for (int y = 0; y < Height; y++)
{
for (int x = 0; x < Width; x++, p++)
{
int predicted = foreseer.Foresee(orig, x, y);
Data[p] = orig.Data[p] - predicted;
if (Data[p] < 0)
{
Data[p] += modulus;
}
foreseer.Learn(orig, x, y, orig.Data[p], predicted);
}
}
}
public override int Foresee(IntField image, int x, int y)
{
_curAreas.Clear();
for (int h = _height; h >= 1; h--)
{
var area = ExtractArea(image, x, y, h);
if (area != null)
{
_curAreas.Add(area);
}
}
foreach (var area in _curAreas)
{
ColorFrequencies cf;
if (_history.TryGetValue(area, out cf))
{
return(cf.Predicted());
}
}
return(_fallback.Foresee(image, x, y));
}