Teaching Computers to Learn•Interactive notebook lesson•Free

cifar10 eda only

No account is required. Learn the material, try the examples, and mark it complete when you are ready to move on. Sign in with email only if you want to save progress across devices.

CIFAR‑10 Exploratory Data Analysis (EDA)

This notebook performs exploratory data analysis on the CIFAR‑10 dataset.

CIFAR‑10 contains 60,000 RGB images belonging to 10 classes.

Image size: 32×32×3 Dataset split:

Training set: 50,000 images
Test set: 10,000 images

Goal of this notebook:

Load and inspect the dataset
Understand image structure
Visualize sample images
Explore class distribution
Analyze pixel values
Prepare data for modeling

⚠️ This notebook focuses only on EDA. Model training is handled separately in the .py script.

Import Libraries

Codepython

import tensorflow as tf
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

Output

2026-03-15 19:15:23.192445: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2026-03-15 19:15:23.214705: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2026-03-15 19:15:23.214724: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2026-03-15 19:15:23.215299: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2026-03-15 19:15:23.219211: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2026-03-15 19:15:23.640358: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT

Load CIFAR‑10 Dataset

Codepython

from tensorflow.keras.datasets import cifar10

(x_train, y_train), (x_test, y_test) = cifar10.load_data()

print("Training images shape:", x_train.shape)
print("Training labels shape:", y_train.shape)

print("Test images shape:", x_test.shape)
print("Test labels shape:", y_test.shape)

Output

Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
170498071/170498071 [==============================] - 20s 0us/step
Training images shape: (50000, 32, 32, 3)
Training labels shape: (50000, 1)
Test images shape: (10000, 32, 32, 3)
Test labels shape: (10000, 1)

Dataset Structure

Each image has shape:

text

32 x 32 x 3

Where:

32 → height
32 → width
3 → RGB color channels

Codepython

x_train[0].shape

Output

(32, 32, 3)

Define Class Labels

Codepython

class_names = [
    "airplane","automobile","bird","cat","deer",
    "dog","frog","horse","ship","truck"
]

class_names

Output

['airplane',
 'automobile',
 'bird',
 'cat',
 'deer',
 'dog',
 'frog',
 'horse',
 'ship',
 'truck']

Visualize Sample Images

Codepython

plt.figure(figsize=(10,5))

for i in range(10):
    plt.subplot(2,5,i+1)
    plt.imshow(x_train[i])
    plt.title(class_names[y_train[i][0]])
    plt.axis("off")

plt.tight_layout()

Output

Class Distribution

Codepython

labels = pd.Series(y_train.flatten())

plt.figure(figsize=(6,4))
sns.countplot(x=labels)
plt.title("Training Label Distribution")
plt.xlabel("Class Label")
plt.ylabel("Count")

Output

Text(0, 0.5, 'Count')

Pixel Value Range

Pixel values range from 0 to 255 for RGB images.

Codepython

print("Minimum pixel value:", x_train.min())
print("Maximum pixel value:", x_train.max())

Output

Minimum pixel value: 0
Maximum pixel value: 255

Pixel Value Distribution

Codepython

plt.figure(figsize=(6,4))
plt.hist(x_train.flatten(), bins=50)
plt.title("Pixel Intensity Distribution")
plt.xlabel("Pixel Value")
plt.ylabel("Frequency")

Output

Text(0, 0.5, 'Frequency')

Normalize Pixel Values

Codepython

x_train_norm = x_train / 255.0
x_test_norm = x_test / 255.0

print("Max normalized value:", np.max(x_train_norm))

Output

Max normalized value: 1.0

Visualize Normalized Image

Codepython

plt.imshow(x_train_norm[0])
plt.title(class_names[y_train[0][0]])
plt.axis("off")

Output

(-0.5, 31.5, 31.5, -0.5)

Mean Image

We compute the average image across the dataset. This helps visualize the overall structure of CIFAR‑10 images.

Codepython

mean_image = np.mean(x_train_norm, axis=0)

plt.imshow(mean_image)
plt.title("Average CIFAR10 Image")
plt.axis("off")

Output

(-0.5, 31.5, 31.5, -0.5)

Conclusion

From this exploratory analysis we learned:

Images are 32×32 RGB
Pixel values range 0–255
Dataset is fairly balanced across classes
Normalization converts pixels to 0–1 range

Next step: Train a CNN model using this dataset (handled in the training script).

Complete cnn basic notebook

Codepython

# ==========================
# 1. Import Libraries
# ==========================
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

from tensorflow.keras.datasets import cifar10
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, BatchNormalization
from sklearn.metrics import classification_report, confusion_matrix

# ==========================
# 2. Load Dataset
# ==========================
(x_train, y_train), (x_test, y_test) = cifar10.load_data()

print("Training images shape:", x_train.shape)
print("Training labels shape:", y_train.shape)
print("Test images shape:", x_test.shape)
print("Test labels shape:", y_test.shape)

# Flatten labels
y_train = y_train.flatten()
y_test = y_test.flatten()

# ==========================
# 3. Normalize Images
# ==========================
x_train = x_train / 255.0
x_test = x_test / 255.0

# ==========================
# 4. Class Names
# ==========================
class_names = [
'airplane','automobile','bird','cat','deer',
'dog','frog','horse','ship','truck'
]

# ==========================
# 5. Build CNN Model
# ==========================
model = Sequential()

# First Convolution Block
model.add(Conv2D(32, (3,3), activation='relu', padding='same', input_shape=(32,32,3)))
model.add(BatchNormalization())
model.add(Conv2D(32, (3,3), activation='relu'))
model.add(MaxPooling2D((2,2)))
model.add(Dropout(0.25))

# Second Convolution Block
model.add(Conv2D(64, (3,3), activation='relu', padding='same'))
model.add(BatchNormalization())
model.add(Conv2D(64, (3,3), activation='relu'))
model.add(MaxPooling2D((2,2)))
model.add(Dropout(0.25))

# Third Convolution Block
model.add(Conv2D(128, (3,3), activation='relu', padding='same'))
model.add(BatchNormalization())
model.add(Conv2D(128, (3,3), activation='relu'))
model.add(MaxPooling2D((2,2)))
model.add(Dropout(0.25))

# Fully Connected Layer
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.5))

# Output Layer
model.add(Dense(10, activation='softmax'))

# ==========================
# 6. Compile Model
# ==========================
model.compile(
    optimizer='adam',
    loss='sparse_categorical_crossentropy',
    metrics=['accuracy']
)

# ==========================
# 7. Model Summary
# ==========================
model.summary()

# ==========================
# 8. Train Model
# ==========================
history = model.fit(
    x_train,
    y_train,
    epochs=20,
    batch_size=64,
    validation_split=0.2
)

# ==========================
# 9. Evaluate Model
# ==========================
test_loss, test_accuracy = model.evaluate(x_test, y_test)

print("\nTest Accuracy:", test_accuracy)

# ==========================
# 10. Predictions
# ==========================
y_pred_probs = model.predict(x_test)
y_pred = np.argmax(y_pred_probs, axis=1)

# ==========================
# 11. Classification Report
# ==========================
print("\nClassification Report:\n")
print(classification_report(y_test, y_pred, target_names=class_names))

# ==========================
# 12. Confusion Matrix
# ==========================
cm = confusion_matrix(y_test, y_pred)

plt.figure(figsize=(10,8))
sns.heatmap(cm, annot=True, fmt="d", cmap="Blues",
            xticklabels=class_names,
            yticklabels=class_names)

plt.xlabel("Predicted")
plt.ylabel("Actual")
plt.title("Confusion Matrix - CIFAR10 CNN")
plt.show()

# ==========================
# 13. Training Curves
# ==========================
plt.figure(figsize=(12,5))

# Accuracy
plt.subplot(1,2,1)
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title("Model Accuracy")
plt.xlabel("Epoch")
plt.ylabel("Accuracy")
plt.legend(["Train","Validation"])

# Loss
plt.subplot(1,2,2)
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title("Model Loss")
plt.xlabel("Epoch")
plt.ylabel("Loss")
plt.legend(["Train","Validation"])

plt.show()

# ==========================
# 14. Show Sample Predictions
# ==========================
plt.figure(figsize=(10,5))

for i in range(10):
    plt.subplot(2,5,i+1)
    plt.imshow(x_test[i])
    plt.title(class_names[y_pred[i]])
    plt.axis("off")

plt.tight_layout()
plt.show()

Output

Training images shape: (50000, 32, 32, 3)
Training labels shape: (50000, 1)
Test images shape: (10000, 32, 32, 3)
Test labels shape: (10000, 1)
Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 conv2d (Conv2D)             (None, 32, 32, 32)        896       
                                                                 
 batch_normalization (Batch  (None, 32, 32, 32)        128       
 Normalization)                                                  
                                                                 
 conv2d_1 (Conv2D)           (None, 30, 30, 32)        9248      
                                                                 
 max_pooling2d (MaxPooling2  (None, 15, 15, 32)        0         
 D)                                                              
                                                                 
 dropout (Dropout)           (None, 15, 15, 32)        0         
                                                                 
 conv2d_2 (Conv2D)           (None, 15, 15, 64)        18496     
                                                                 
 batch_normalization_1 (Bat  (None, 15, 15, 64)        256       
 chNormalization)                                                
                                                                 
 conv2d_3 (Conv2D)           (None, 13, 13, 64)        36928     
                                                                 
 max_pooling2d_1 (MaxPoolin  (None, 6, 6, 64)          0         
 g2D)                                                            
                                                                 
 dropout_1 (Dropout)         (None, 6, 6, 64)          0         
                                                                 
 conv2d_4 (Conv2D)           (None, 6, 6, 128)         73856     
                                                                 
 batch_normalization_2 (Bat  (None, 6, 6, 128)         512       
 chNormalization)                                                
                                                                 
 conv2d_5 (Conv2D)           (None, 4, 4, 128)         147584    
                                                                 
 max_pooling2d_2 (MaxPoolin  (None, 2, 2, 128)         0         
 g2D)                                                            
                                                                 
 dropout_2 (Dropout)         (None, 2, 2, 128)         0         
                                                                 
 flatten (Flatten)           (None, 512)               0         
                                                                 
 dense (Dense)               (None, 256)               131328    
                                                                 
 dropout_3 (Dropout)         (None, 256)               0         
                                                                 
 dense_1 (Dense)             (None, 10)                2570      
                                                                 
=================================================================
Total params: 421802 (1.61 MB)
Trainable params: 421354 (1.61 MB)
Non-trainable params: 448 (1.75 KB)
_________________________________________________________________

2026-03-15 19:15:50.461197: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.483936: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.487565: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.490735: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.494331: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.496803: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.570755: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.571629: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.572373: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2026-03-15 19:15:50.573099: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1929] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 1181 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 3060, pci bus id: 0000:01:00.0, compute capability: 8.6

Epoch 1/20

2026-03-15 19:15:51.788492: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:961] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential/dropout/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer
2026-03-15 19:15:51.913189: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:454] Loaded cuDNN version 8900
2026-03-15 19:15:52.724093: I external/local_xla/xla/service/service.cc:168] XLA service 0x78e5d3833a60 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2026-03-15 19:15:52.724105: I external/local_xla/xla/service/service.cc:176]   StreamExecutor device (0): NVIDIA GeForce RTX 3060, Compute Capability 8.6
2026-03-15 19:15:52.727613: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1773602152.793138  203003 device_compiler.h:186] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

625/625 [==============================] - 7s 6ms/step - loss: 1.6596 - accuracy: 0.3954 - val_loss: 1.4247 - val_accuracy: 0.4699
Epoch 2/20
625/625 [==============================] - 3s 5ms/step - loss: 1.2530 - accuracy: 0.5534 - val_loss: 1.1891 - val_accuracy: 0.5769
Epoch 3/20
625/625 [==============================] - 3s 5ms/step - loss: 1.0553 - accuracy: 0.6318 - val_loss: 1.0292 - val_accuracy: 0.6362
Epoch 4/20
625/625 [==============================] - 3s 5ms/step - loss: 0.9316 - accuracy: 0.6758 - val_loss: 1.2344 - val_accuracy: 0.6089
Epoch 5/20
625/625 [==============================] - 3s 5ms/step - loss: 0.8550 - accuracy: 0.7043 - val_loss: 0.9215 - val_accuracy: 0.6829
Epoch 6/20
625/625 [==============================] - 3s 5ms/step - loss: 0.7843 - accuracy: 0.7269 - val_loss: 0.7217 - val_accuracy: 0.7511
Epoch 7/20
625/625 [==============================] - 3s 5ms/step - loss: 0.7389 - accuracy: 0.7470 - val_loss: 0.6661 - val_accuracy: 0.7698
Epoch 8/20
625/625 [==============================] - 3s 5ms/step - loss: 0.7061 - accuracy: 0.7586 - val_loss: 0.6754 - val_accuracy: 0.7688
Epoch 9/20
625/625 [==============================] - 3s 5ms/step - loss: 0.6646 - accuracy: 0.7738 - val_loss: 0.6782 - val_accuracy: 0.7711
Epoch 10/20
625/625 [==============================] - 3s 5ms/step - loss: 0.6342 - accuracy: 0.7820 - val_loss: 0.7003 - val_accuracy: 0.7646
Epoch 11/20
625/625 [==============================] - 3s 5ms/step - loss: 0.6070 - accuracy: 0.7916 - val_loss: 0.6890 - val_accuracy: 0.7706
Epoch 12/20
625/625 [==============================] - 3s 5ms/step - loss: 0.5894 - accuracy: 0.7997 - val_loss: 0.7064 - val_accuracy: 0.7598
Epoch 13/20
625/625 [==============================] - 3s 5ms/step - loss: 0.5622 - accuracy: 0.8066 - val_loss: 0.6222 - val_accuracy: 0.7929
Epoch 14/20
625/625 [==============================] - 3s 5ms/step - loss: 0.5438 - accuracy: 0.8139 - val_loss: 0.5839 - val_accuracy: 0.8070
Epoch 15/20
625/625 [==============================] - 3s 5ms/step - loss: 0.5226 - accuracy: 0.8196 - val_loss: 0.6641 - val_accuracy: 0.7850
Epoch 16/20
625/625 [==============================] - 3s 5ms/step - loss: 0.5131 - accuracy: 0.8230 - val_loss: 0.6149 - val_accuracy: 0.7977
Epoch 17/20
625/625 [==============================] - 3s 5ms/step - loss: 0.4956 - accuracy: 0.8288 - val_loss: 0.5942 - val_accuracy: 0.7969
Epoch 18/20
625/625 [==============================] - 3s 5ms/step - loss: 0.4824 - accuracy: 0.8329 - val_loss: 0.5808 - val_accuracy: 0.8143
Epoch 19/20
625/625 [==============================] - 3s 5ms/step - loss: 0.4681 - accuracy: 0.8389 - val_loss: 0.5745 - val_accuracy: 0.8104
Epoch 20/20
625/625 [==============================] - 3s 5ms/step - loss: 0.4582 - accuracy: 0.8426 - val_loss: 0.5418 - val_accuracy: 0.8226
313/313 [==============================] - 1s 1ms/step - loss: 0.5620 - accuracy: 0.8157

Test Accuracy: 0.8156999945640564
313/313 [==============================] - 0s 966us/step

Classification Report:

              precision    recall  f1-score   support

    airplane       0.81      0.85      0.83      1000
  automobile       0.92      0.92      0.92      1000
        bird       0.85      0.66      0.74      1000
         cat       0.59      0.73      0.65      1000
        deer       0.80      0.78      0.79      1000
         dog       0.73      0.72      0.73      1000
        frog       0.83      0.87      0.85      1000
       horse       0.85      0.84      0.85      1000
        ship       0.92      0.89      0.91      1000
       truck       0.91      0.88      0.89      1000

    accuracy                           0.82     10000
   macro avg       0.82      0.82      0.82     10000
weighted avg       0.82      0.82      0.82     10000

[1;31mThe Kernel crashed while executing code in the current cell or a previous cell. 
[1;31mPlease review the code in the cell(s) to identify a possible cause of the failure. 
[1;31mClick <a href='https://aka.ms/vscodeJupyterKernelCrash'>here</a> for more info. 
[1;31mView Jupyter <a href='command:jupyter.viewOutput'>log</a> for further details.

Lesson resources

Supporting code and files from this part of the curriculum.

mnistpy train cifar10 cnnpy cats vs dogs vgg16 fullpy apppy MNIST without one hot vectorspdf

Finished this lesson?

Keep your place here, or save it across devices with an email link.

Saved in this browser. Save progress with email →

CIFAR‑10 Exploratory Data Analysis (EDA)

This notebook performs exploratory data analysis on the CIFAR‑10 dataset.

CIFAR‑10 contains 60,000 RGB images belonging to 10 classes.

Image size: 32×32×3 Dataset split:

Training set: 50,000 images

Test set: 10,000 images

Goal of this notebook:

Load and inspect the dataset

Understand image structure

Visualize sample images

Explore class distribution

Analyze pixel values

Prepare data for modeling

⚠️ This notebook focuses only on EDA. Model training is handled separately in the .py script.

2026-03-15 19:15:23.192445: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. 2026-03-15 19:15:23.214705: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered 2026-03-15 19:15:23.214724: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered 2026-03-15 19:15:23.215299: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered 2026-03-15 19:15:23.219211: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations. To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags. 2026-03-15 19:15:23.640358: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT

from tensorflow.keras.datasets import cifar10 (x_train, y_train), (x_test, y_test) = cifar10.load_data() print("Training images shape:", x_train.shape) print("Training labels shape:", y_train.shape) print("Test images shape:", x_test.shape) print("Test labels shape:", y_test.shape)

Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz 170498071/170498071 [==============================] - 20s 0us/step Training images shape: (50000, 32, 32, 3) Training labels shape: (50000, 1) Test images shape: (10000, 32, 32, 3) Test labels shape: (10000, 1)

# ========================== # 1. Import Libraries # ========================== import tensorflow as tf import numpy as np import matplotlib.pyplot as plt import seaborn as sns from tensorflow.keras.datasets import cifar10 from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, BatchNormalization from sklearn.metrics import classification_report, confusion_matrix # ========================== # 2. Load Dataset # ========================== (x_train, y_train), (x_test, y_test) = cifar10.load_data() print("Training images shape:", x_train.shape) print("Training labels shape:", y_train.shape) print("Test images shape:", x_test.shape) print("Test labels shape:", y_test.shape) # Flatten labels y_train = y_train.flatten() y_test = y_test.flatten() # ========================== # 3. Normalize Images # ========================== x_train = x_train / 255.0 x_test = x_test / 255.0 # ========================== # 4. Class Names # ========================== class_names = [ 'airplane','automobile','bird','cat','deer', 'dog','frog','horse','ship','truck' ] # ========================== # 5. Build CNN Model # ========================== model = Sequential() # First Convolution Block model.add(Conv2D(32, (3,3), activation='relu', padding='same', input_shape=(32,32,3))) model.add(BatchNormalization()) model.add(Conv2D(32, (3,3), activation='relu')) model.add(MaxPooling2D((2,2))) model.add(Dropout(0.25)) # Second Convolution Block model.add(Conv2D(64, (3,3), activation='relu', padding='same')) model.add(BatchNormalization()) model.add(Conv2D(64, (3,3), activation='relu')) model.add(MaxPooling2D((2,2))) model.add(Dropout(0.25)) # Third Convolution Block model.add(Conv2D(128, (3,3), activation='relu', padding='same')) model.add(BatchNormalization()) model.add(Conv2D(128, (3,3), activation='relu')) model.add(MaxPooling2D((2,2))) model.add(Dropout(0.25)) # Fully Connected Layer model.add(Flatten()) model.add(Dense(256, activation='relu')) model.add(Dropout(0.5)) # Output Layer model.add(Dense(10, activation='softmax')) # ========================== # 6. Compile Model # ========================== model.compile( optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'] ) # ========================== # 7. Model Summary # ========================== model.summary() # ========================== # 8. Train Model # ========================== history = model.fit( x_train, y_train, epochs=20, batch_size=64, validation_split=0.2 ) # ========================== # 9. Evaluate Model # ========================== test_loss, test_accuracy = model.evaluate(x_test, y_test) print("\nTest Accuracy:", test_accuracy) # ========================== # 10. Predictions # ========================== y_pred_probs = model.predict(x_test) y_pred = np.argmax(y_pred_probs, axis=1) # ========================== # 11. Classification Report # ========================== print("\nClassification Report:\n") print(classification_report(y_test, y_pred, target_names=class_names)) # ========================== # 12. Confusion Matrix # ========================== cm = confusion_matrix(y_test, y_pred) plt.figure(figsize=(10,8)) sns.heatmap(cm, annot=True, fmt="d", cmap="Blues", xticklabels=class_names, yticklabels=class_names) plt.xlabel("Predicted") plt.ylabel("Actual") plt.title("Confusion Matrix - CIFAR10 CNN") plt.show() # ========================== # 13. Training Curves # ========================== plt.figure(figsize=(12,5)) # Accuracy plt.subplot(1,2,1) plt.plot(history.history['accuracy']) plt.plot(history.history['val_accuracy']) plt.title("Model Accuracy") plt.xlabel("Epoch") plt.ylabel("Accuracy") plt.legend(["Train","Validation"]) # Loss plt.subplot(1,2,2) plt.plot(history.history['loss']) plt.plot(history.history['val_loss']) plt.title("Model Loss") plt.xlabel("Epoch") plt.ylabel("Loss") plt.legend(["Train","Validation"]) plt.show() # ========================== # 14. Show Sample Predictions # ========================== plt.figure(figsize=(10,5)) for i in range(10): plt.subplot(2,5,i+1) plt.imshow(x_test[i]) plt.title(class_names[y_pred[i]]) plt.axis("off") plt.tight_layout() plt.show()

Training images shape: (50000, 32, 32, 3) Training labels shape: (50000, 1) Test images shape: (10000, 32, 32, 3) Test labels shape: (10000, 1) Model: "sequential" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= conv2d (Conv2D) (None, 32, 32, 32) 896 batch_normalization (Batch (None, 32, 32, 32) 128 Normalization) conv2d_1 (Conv2D) (None, 30, 30, 32) 9248 max_pooling2d (MaxPooling2 (None, 15, 15, 32) 0 D) dropout (Dropout) (None, 15, 15, 32) 0 conv2d_2 (Conv2D) (None, 15, 15, 64) 18496 batch_normalization_1 (Bat (None, 15, 15, 64) 256 chNormalization) conv2d_3 (Conv2D) (None, 13, 13, 64) 36928 max_pooling2d_1 (MaxPoolin (None, 6, 6, 64) 0 g2D) dropout_1 (Dropout) (None, 6, 6, 64) 0 conv2d_4 (Conv2D) (None, 6, 6, 128) 73856 batch_normalization_2 (Bat (None, 6, 6, 128) 512 chNormalization) conv2d_5 (Conv2D) (None, 4, 4, 128) 147584 max_pooling2d_2 (MaxPoolin (None, 2, 2, 128) 0 g2D) dropout_2 (Dropout) (None, 2, 2, 128) 0 flatten (Flatten) (None, 512) 0 dense (Dense) (None, 256) 131328 dropout_3 (Dropout) (None, 256) 0 dense_1 (Dense) (None, 10) 2570 ================================================================= Total params: 421802 (1.61 MB) Trainable params: 421354 (1.61 MB) Non-trainable params: 448 (1.75 KB) _________________________________________________________________

2026-03-15 19:15:50.461197: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.483936: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.487565: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.490735: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.494331: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.496803: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.570755: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.571629: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.572373: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355 2026-03-15 19:15:50.573099: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1929] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 1181 MB memory: -> device: 0, name: NVIDIA GeForce RTX 3060, pci bus id: 0000:01:00.0, compute capability: 8.6

2026-03-15 19:15:51.788492: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:961] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential/dropout/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer 2026-03-15 19:15:51.913189: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:454] Loaded cuDNN version 8900 2026-03-15 19:15:52.724093: I external/local_xla/xla/service/service.cc:168] XLA service 0x78e5d3833a60 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices: 2026-03-15 19:15:52.724105: I external/local_xla/xla/service/service.cc:176] StreamExecutor device (0): NVIDIA GeForce RTX 3060, Compute Capability 8.6 2026-03-15 19:15:52.727613: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable. WARNING: All log messages before absl::InitializeLog() is called are written to STDERR I0000 00:00:1773602152.793138 203003 device_compiler.h:186] Compiled cluster using XLA! This line is logged at most once for the lifetime of the process.

625/625 [==============================] - 7s 6ms/step - loss: 1.6596 - accuracy: 0.3954 - val_loss: 1.4247 - val_accuracy: 0.4699 Epoch 2/20 625/625 [==============================] - 3s 5ms/step - loss: 1.2530 - accuracy: 0.5534 - val_loss: 1.1891 - val_accuracy: 0.5769 Epoch 3/20 625/625 [==============================] - 3s 5ms/step - loss: 1.0553 - accuracy: 0.6318 - val_loss: 1.0292 - val_accuracy: 0.6362 Epoch 4/20 625/625 [==============================] - 3s 5ms/step - loss: 0.9316 - accuracy: 0.6758 - val_loss: 1.2344 - val_accuracy: 0.6089 Epoch 5/20 625/625 [==============================] - 3s 5ms/step - loss: 0.8550 - accuracy: 0.7043 - val_loss: 0.9215 - val_accuracy: 0.6829 Epoch 6/20 625/625 [==============================] - 3s 5ms/step - loss: 0.7843 - accuracy: 0.7269 - val_loss: 0.7217 - val_accuracy: 0.7511 Epoch 7/20 625/625 [==============================] - 3s 5ms/step - loss: 0.7389 - accuracy: 0.7470 - val_loss: 0.6661 - val_accuracy: 0.7698 Epoch 8/20 625/625 [==============================] - 3s 5ms/step - loss: 0.7061 - accuracy: 0.7586 - val_loss: 0.6754 - val_accuracy: 0.7688 Epoch 9/20 625/625 [==============================] - 3s 5ms/step - loss: 0.6646 - accuracy: 0.7738 - val_loss: 0.6782 - val_accuracy: 0.7711 Epoch 10/20 625/625 [==============================] - 3s 5ms/step - loss: 0.6342 - accuracy: 0.7820 - val_loss: 0.7003 - val_accuracy: 0.7646 Epoch 11/20 625/625 [==============================] - 3s 5ms/step - loss: 0.6070 - accuracy: 0.7916 - val_loss: 0.6890 - val_accuracy: 0.7706 Epoch 12/20 625/625 [==============================] - 3s 5ms/step - loss: 0.5894 - accuracy: 0.7997 - val_loss: 0.7064 - val_accuracy: 0.7598 Epoch 13/20 625/625 [==============================] - 3s 5ms/step - loss: 0.5622 - accuracy: 0.8066 - val_loss: 0.6222 - val_accuracy: 0.7929 Epoch 14/20 625/625 [==============================] - 3s 5ms/step - loss: 0.5438 - accuracy: 0.8139 - val_loss: 0.5839 - val_accuracy: 0.8070 Epoch 15/20 625/625 [==============================] - 3s 5ms/step - loss: 0.5226 - accuracy: 0.8196 - val_loss: 0.6641 - val_accuracy: 0.7850 Epoch 16/20 625/625 [==============================] - 3s 5ms/step - loss: 0.5131 - accuracy: 0.8230 - val_loss: 0.6149 - val_accuracy: 0.7977 Epoch 17/20 625/625 [==============================] - 3s 5ms/step - loss: 0.4956 - accuracy: 0.8288 - val_loss: 0.5942 - val_accuracy: 0.7969 Epoch 18/20 625/625 [==============================] - 3s 5ms/step - loss: 0.4824 - accuracy: 0.8329 - val_loss: 0.5808 - val_accuracy: 0.8143 Epoch 19/20 625/625 [==============================] - 3s 5ms/step - loss: 0.4681 - accuracy: 0.8389 - val_loss: 0.5745 - val_accuracy: 0.8104 Epoch 20/20 625/625 [==============================] - 3s 5ms/step - loss: 0.4582 - accuracy: 0.8426 - val_loss: 0.5418 - val_accuracy: 0.8226 313/313 [==============================] - 1s 1ms/step - loss: 0.5620 - accuracy: 0.8157 Test Accuracy: 0.8156999945640564 313/313 [==============================] - 0s 966us/step Classification Report: precision recall f1-score support airplane 0.81 0.85 0.83 1000 automobile 0.92 0.92 0.92 1000 bird 0.85 0.66 0.74 1000 cat 0.59 0.73 0.65 1000 deer 0.80 0.78 0.79 1000 dog 0.73 0.72 0.73 1000 frog 0.83 0.87 0.85 1000 horse 0.85 0.84 0.85 1000 ship 0.92 0.89 0.91 1000 truck 0.91 0.88 0.89 1000 accuracy 0.82 10000 macro avg 0.82 0.82 0.82 10000 weighted avg 0.82 0.82 0.82 10000

[1;31mThe Kernel crashed while executing code in the current cell or a previous cell. [1;31mPlease review the code in the cell(s) to identify a possible cause of the failure. [1;31mClick <a href='https://aka.ms/vscodeJupyterKernelCrash'>here</a> for more info. [1;31mView Jupyter <a href='command:jupyter.viewOutput'>log</a> for further details.