Fix rand function (instead of generating integral and fractional, generate a number in [0, MAX_INT] and multiply by max); adjust tests to match

Author: djslzx

mnist-clojure.iml

@@ -6,8 +6,8 @@
     <exclude-output />
     <content url="file://$MODULE_DIR$">
       <sourceFolder url="file://$MODULE_DIR$/resources" isTestSource="false" />
-      <sourceFolder url="file://$MODULE_DIR$/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/dev-resources" isTestSource="false" />
+      <sourceFolder url="file://$MODULE_DIR$/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/test" isTestSource="true" />
       <excludeFolder url="file://$MODULE_DIR$/target/default" />
     </content>

src/mnist_clojure/core.clj

@@ -30,8 +30,9 @@
     (+ 256 byte)
     byte))
 
-(defn read-mnist-file [file-name]
+(defn read-mnist-file
   "Reads the idx[x]-ubyte format and parses it into a byte array"
+  [file-name]
   (let [file (io/file file-name)
         b-array (byte-array (.length file))]
     (with-open [stream (io/input-stream file)]
@@ -104,8 +105,9 @@
   "Creates a list of specified length with random decimal values between 0 to max"
   {:test (fn []
            (is= (rand-d-list 4 1 0.4)
-                [1961823115700386051
-                 '(0.1430590959000676 0.8178222714074991 0.5044600700514671 0.01660157088963388)]))}
+                [-9136436700791295257
+                 [0.20178402802058684 0.02002609726974093
+                  0.006640628355853552 1.8626451500983188E-10]]))}
   [length seed max]
   (reduce (fn [[seed rand-vals] _]
             (let [[next-seed rand-val] (rand seed max)]
@@ -119,7 +121,10 @@
            (is= (rand-list-max-mag 4 2 0)
                 [-9197343212719499864 `(0 0 0 0)])
            (is= (rand-list-max-mag 3 4 23)
-                [-7068052242903947273 `(-6.004336956424794 13.63314510538855 -18.93359371783845)]))}
+                [77986490623247743
+                 [-22.18838978334721
+                  -19.945311020568624
+                  -22.999999914318323]]))}
   [length seed max]
   (let [[new-seed rand-list] (rand-d-list length seed (* 2 max))]
     [new-seed (map (partial + (- max)) rand-list)]))
@@ -128,11 +133,18 @@
   "Initializes one layer of the neural network. Note the + 1 adjusts for the bias"
   ; y = Wx
   {:test (fn []
-           (is= (initialize-layer 1 3 1 0) [-8728512804673154413 `((0 0) (0 0) (0 0))])
+           (is= (initialize-layer 1 3 1 0)
+                [-8728512804673154413 [[0 0] [0 0] [0 0]]])
            (is= (initialize-layer 3 2 24 1)
-                [7786394753034826687
-                 '((-0.32256568051994106 0.5576858765248609 0.5806762038640101 -0.8371181152002505)
-                   (0.6701701863995613 0.3397951933274954 -0.48321265703216776 -0.6015623093589966))]))}
+                [2665986749794895764
+                 [[0.34034037279912277
+                   -0.24479991162419323
+                   -0.32040961334500895
+                   -0.38945634541542096]
+                  [0.03357468593566448
+                   -0.9542951490517217
+                   -0.20312461871799303
+                   -0.9999999776482582]]]))}
   [input-count output-count seed max]
   (reduce (fn [[seed rand-lists] _]
             (let [[new-seed rand-list] (rand-list-max-mag (+ input-count 1) seed max)]
@@ -246,6 +258,6 @@
                               "resources/train-labels.idx1-ubyte" 0.2 10 1))))
 
 (comment (time (train-and-test "resources/train-images.idx3-ubyte"
-                         "resources/train-labels.idx1-ubyte"
-                         "resources/test-images.idx3-ubyte"
-                         "resources/test-labels.idx1-ubyte")))
+                               "resources/train-labels.idx1-ubyte"
+                               "resources/test-images.idx3-ubyte"
+                               "resources/test-labels.idx1-ubyte")))

src/mnist_clojure/random.clj

@@ -3,34 +3,40 @@
             [ysera.test :refer [is is= is-not]])
   (:refer-clojure :exclude [rand rand-int]))
 
+(defn get-new-seed
+  "Returns a new seed"
+  {:test (fn []
+           (is= (get-new-seed 0) 1)
+           (is= (get-new-seed 1) 35651602))}
+  [seed]
+  (let [[new-seed _] (get-random-int seed 1)]
+    new-seed))
+
 (defn rand-int
   "Returns [new-seed random-int] such that 0 <= rand-int < max."
   {:test (fn []
+           (is= (rand-int 1 1) [35651602 0])
+           (is= (rand-int 1 0.1) [35651602 0])
            (is= (rand-int 1 3) [35651602 1])
-           (is= (rand-int 3158 13984) [108024351031 3158])
            ; rand-int always returns 0 without errors no matter the seed.
            (is= (rand-int 3 0) [106954804 0])
-           (is= (rand-int 1513123 0) [49270036527376 0]))}
+           (is= (rand-int -3 0) [94371886 0])
+           )}
   [seed max]
   (if (< max 1)
-    (let [[new-seed _] (get-random-int seed 1)]
-      [new-seed 0])
+    [(get-new-seed seed) 0]
     (get-random-int seed max)))
 
 (defn rand
   "Returns [new-seed rand] such that 0 <= rand < max."
   {:test (fn []
-           (is= (rand 3 4) [3390894109721333 3.049804711737579])
-           (is= (rand 1023193 9) [5555018649264297881 1.3268604484977482])
-           (is= (rand 1319 0) [46890243928 0])
-           (is= (rand 19220393 0) [684066761437437 0]))}
+           (is= (rand 1 0) [35651602 0])
+           (is= (rand 1 1) [35651602 4.656612875245797E-10])
+           (is= (rand 1 0.5) [35651602 2.3283064376228985E-10])
+           (is= (rand 1 0.1) [35651602 4.656612875245797E-11]))}
   [seed max]
-  (if (= max 0)
-    (rand-int seed max)
-    (let [[new-seed random-int] (rand-int seed max)
-          [nn-seed new-random-int] (rand-int new-seed Integer/MAX_VALUE)]
-      (if (= new-random-int 0)
-        [nn-seed random-int]
-            [nn-seed (+ random-int (double (/ new-random-int Integer/MAX_VALUE)))]))))
-
-
+  (if (<= max 0)
+    [(get-new-seed seed) 0]
+    (let [[seed n] (rand-int seed Integer/MAX_VALUE)
+          weight (double (/ n Integer/MAX_VALUE))]
+      [seed (* max weight)])))