The paper you link to includes attacks on blockchain applications as "attacks on blockchain". Personally I would restrict "attacks on blockchain" to literal attacks on the underlying blockchain rather than applications built on top. Regardless there are a number of theoretical attacks on blockchains (51 percent, selfish mining etc) but I don't think we have any evidence of them being executed on Bitcoin at least.
Selfish mining is when miners collude to withhold newly mined blocks and start mining on top of them without announcing the newly mined block to the network. It is not an attack on the blockchain per se, rather than the profitability of other honest miners. In this case honest miners may mine a block that ends up getting orphaned because they were unaware of the latest blockchain tip.
A 51 percent attack on the other hand is most definitely an attack on a blockchain. If a miner tried to execute a 51 percent attack (with 51 percent of the network hashrate) they could theoretically include a double spend in the chain with the most accumulated proof of work. In practice the attack may fail as full nodes on the network reject the chain with the double spend in a block and fork to a chain that doesn't include that block. This resource from MIT highlights some examples of attempted 51 percent attacks on blockchains other than Bitcoin.