TJCTF 2020 - Jarvis
Looking at the files we are given, we realized that the challenge is a binary classification task. A simple search reveals this script which we adapted to the help.csv dataset.
from numpy import loadtxt
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
help_dataset = loadtxt("help.csv", delimiter=",")
x = help_dataset[:, 1:10]
y = help_dataset[:, 0]
model = Sequential()
model.add(Dense(12, input_dim=9, activation="relu"))
model.add(Dense(8, activation="relu"))
model.add(Dense(1, activation="sigmoid"))
model.compile(loss="binary_crossentropy", optimizer="adam", metrics=["accuracy"])
model.fit(x, y, epochs=50, batch_size=10)
flag_dataset = loadtxt("flag.csv", delimiter=",")
x = flag_dataset[:, 0:9]
predictions = (model.predict(x) > 0.5).astype("int32")
flag = "".join([str(round(float(prediction))) for prediction in predictions])
print(flag)
We get flag{mnWis_cool} and guess the correct flag.
flag{ml_is_cool}