update specification

Author: Cheng-Lin-Li

FastMap/README.md

@@ -9,6 +9,11 @@ The first two columns in each line of the data file represent the IDs of the two
 
 The objects listed in fastmap-data.txt are actually the words in fastmap-wordlist.txt (nth word in this list has an ID value of n) and the distances between each pair of objects are the Damerau–Levenshtein distances between them. The program will plot the words onto a 2D plane using the previous FastMap solution and see what it looks like.
 
+## Technical Specification and Report
+
+Click ** [Here](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/FastMap/TechnicalSpecification-%5BPCA_FastMap%5D-%5B1.0%5D-%5B20160929%5D.pdf) ** to read the detail specification and report.
+
+
 #### Usage: python FastMap.py
 
 

GMM/README.md

@@ -9,6 +9,10 @@ I based on his implementation with modifications as this result.
 Run the algorithm on the data file "clusters.txt" using K, the number of clusters, set to 3. 
 Report the mean (centroids), amplitude and covariance matrix of each Gaussian in GMM
 
+## Technical Specification and Report
+
+Click [** Here **](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/GMM/INF552-TechnicalSpecification-%5Bk-means_EM-GMM%5D-%5B1.2%5D-%5B20170515%5D.pdf) to read the detail specification and report.
+
 #### Usage: python GMM.py	
 
 #### Input: A data file (clusters.txt) that contains 150 2D points. Each row in the file contains the coordinates of a single point.

HMM/README.md

@@ -8,6 +8,10 @@ The file hmm-data.txt contains a map of a 10-by-10 2D grid-world. The most up-le
 
 At a given cell, the robot measures L2 distances (Euclidean distances) to each of the towers. For a true distance d, the robot records a noisy measurement chosen uniformly at random from the set of numbers in the interval [0.7d, 1.3d] with one decimal place. These measurements for 11 time-steps are also provided in the data file. You should output the coordinates of the most likely trajectory of the robot for 11 time-steps. The Viterbi algorithm is the implemented algorithm for this task. For tie breaking, it always prefer the one with a smaller x coordinate, and a smaller y coordinate if the x coordinates are equal.
 
+## Technical Specification and Report
+
+Click [** Here **](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/HMM/INF552-TechnicalSpecification-%5BHMM%5D-%5B1.0%5D-%5B20161203%5D.pdf) to read the detail specification and report.
+
 
 #### Usage: python DecisionTree.py dt-data.txt	
 

NeuralNetwork/README.md

@@ -14,6 +14,9 @@ The label of an image is 1 if the word "down" is in its file name; otherwise the
 
 The image file format is "pgm" <http://netpbm.sourceforge.net/doc/pgm.html>. You may follow the link for the format details. 
 
+## Technical Specification and Report
+
+Click [** Here **](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/NeuralNetwork/INF552-TechnicalSpecification-%5BNeuralNetwork%5D-%5B1.0%5D-%5B20161104%5D.pdf) to read the detail specification and report.
 
 ## Objectives:
 Implement the Back Propagation algorithm for Feed Forward Neural Networks to learn down gestures from training images available in downgesture_train.list. After the training, use the trained network to predict the labels for the gestures in the test images available in downgesture_test.list.

TensorFlow/README.md

@@ -3,6 +3,8 @@
 ## Project:
 The one of the most challenging issue is stock price or indices prediction in the financial industry. On the other hand, machine learning and big data techniques in vision recognition has matured considerably over the last decade. This research adopts Multi-Layer Perceptron (MLP) and Long-Short Term Memory (LSTMs) neural networks to compete with Dynamic-radius Species-conserving Genetic Algorithm (DSGA) for short term stock price prediction. The result indicates that MLP may have a better potential than DSGA on short term stock price prediction and that LSTMs may require more training data to surpass DSGA. 
 
+## Technical Specification and Report
+ Click [**Here**](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/TensorFlow/ProjectReport.pdf) to read the report. 
 
 ### Usage: python StockPriceForecasting.py (or StockPriceForecasting-LSTM.py)	
 
@@ -12,7 +14,7 @@ The text file include 750 data records.  360 are from the first quarter of the y
 ### Output: Iterations and training / testing data results of selected stock index prediction.
 
 
-### Report: Click [**Here**](https://github.com/Cheng-Lin-Li/MachineLearning/blob/master/TensorFlow/ProjectReport.pdf) to read the report. 
+
 
 ### Warning: