 In this video, I will give an overview of our paper optimizing rectangle attacks for unified and generic framework for K recovery. Rectangle attack is a chosen plane test variant of Bromero attack, which is a differential steel attack using two short differentials of high probability to construct a long one. To mount rectangle K recovery attacks as efficiently as possible, four rectangle K recovery algorithms have been presented using four different kinds of sub-K guessing strategies. However, they are treated and separated once before. In our paper, we investigate the rectangle K recovery in depth and come up with a unified and generic rectangle K recovery algorithm supporting any possible attacking parameters, which covers all the previous rectangle K recovery algorithms and unveils another five types of new attacks. Our basic idea is that any possible K guessing strategy should be allowed and that there must be a guessing strategy leading to optimal complexities of the K recovery attack. As a compliment, we propose a framework to tell the best attacking parameters, including the sub-K bits to be guessed. Our automatic framework is modular and flexible. It takes the sets of parameters of a given Bromero or rectangle distinguisher as input and returns the optimal parameters and the time complexity. When we feed the parameters returned by this work to our K recovery algorithm, the time complexity of the attack will be minimal. To demonstrate the efficiency of our new K recovery algorithm, we apply it to four block ciphers using existing distinguishes and obtain a series of improved results. We give better attacks on turn round serpent, improve some of the best attacks for skinny in previous works, extend the rectangle attack on craft by one round and give the first 19 round attack and improve the 11 round rectangle attack and extend the Bromero attack by one round in the related 2K setting for deoxys. Our results are summarized in the table and the results of this work are highlighted in the table. According to these applications, we find that the best attacking parameters differ significantly from those which were used in previous works and even the number of rounds added around the distinguisher is different. It is likely that previous rectangle attacks can be improved to some extent using our new K recovery algorithm. Besides, we develop variants of the new K recovery algorithm for related attacks, including the case rb equals n, the variants for Bromero attacks and the related K setting. That's all. Thank you.