Karlhigley LSH人工神經網絡模型尋找最近的鄰居給零結果

Question

我想找到最接近我的每個點的鄰居，我嘗試了使用karlhigley神經網絡模型。這是一段代碼

List<Tuple2<Object, SparseVector>> svList = new ArrayList<>(); 
     svList.add(new Tuple2<Object, SparseVector>(3L, 
       (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 }, 
         new double[] { 5.0f, 3.0f, 4.0f, 5.0f, 5.0f, 1.0f, 5.0f, 3.0f, 4.0f, 5.0f, 5.0f, 3.0f, 4.0f, 
           5.0f, 5.0f, 1.0f, 5.0f, 3.0f, 4.0f, 5.0f }) 
         .toSparse()))); 
     svList.add(new Tuple2<Object, SparseVector>(4L, 
       (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 }, 
         new double[] { 1.0f, 2.0f, 1.0f, 5.0f, 1.0f, 5.0f, 1.0f, 4.0f, 1.0f, 3.0, 1.0f, 2.0f, 1.0f, 
           5.0f, 1.0f, 5.0f, 1.0f, 4.0f, 1.0f, 3.0f }) 
         .toSparse()))); 
     svList.add(new Tuple2<Object, SparseVector>(6L, 
       (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 }, 
         new double[] { 5.0f, 3.0f, 4.0f, 1.0f, 5.0f, 4.0f, 1.0f, 3.0f, 4.0f, 5.0f, 5.0f, 3.0f, 4.0f, 
           1.0f, 5.0f, 4.0f, 1.0f, 3.0f, 4.0f, 5.0f }) 
         .toSparse()))); 

RDD<Tuple2<Object, SparseVector>> points = sc.parallelize(svList).rdd(); 

ANNModel annModel = 
       new ANN(20, "cosine") 
       .setTables(1) 
       .setSignatureLength(20).setRandomSeed(3) 
       .train(points,StorageLevel.MEMORY_AND_DISK()); 

JavaRDD<Tuple2<Object, Tuple2<Object, Object>[]>> neighbors2 = annModel.neighbors(3).toJavaRDD(); 

JavaRDD neighbors2給我所有的鄰居和他們的分數爲空。任何人都可以幫助我理解我在哪裏實施錯誤，以及如何以正確的方式做到這一點？

這是怎麼了打印輸出

neighbors2.foreach(f->{ 
      for(int i=0;i<f._2.length;i++){ 
       System.out.println(f._1+"====="+f._2[i]._1+"---"+f._2[i]._2); 
      } 
     }); 

Answer 1

解決方案：

print語句是錯誤的。它必須是：

System.out.println(f._1()+"====="+f._2()[i]._1()+"---"+f._2()[i]._2()); 

原因： 由於f._2和f._2之間的類型轉換問題（）。 這篇文章將詳細解釋它： Java Tuple2 difference between using accessor method and calling the variable directly