//#pragma GCC target("avx")
//#pragma GCC optimize("O3")
//#pragma GCC optimize("unroll-loops")
#include<bits/stdc++.h>

#ifdef LOCAL
#include <debug.hpp>
#define debug(...) debug_print::multi_print(#__VA_ARGS__, __VA_ARGS__)
#else
#define debug(...) (static_cast<void>(0))
#endif
using namespace std;
using ll = long long;
using ld = long double;
using pll = pair<ll, ll>;
using pii = pair<int, int>;
using vi = vector<int>;
using vvi = vector<vi>;
using vvvi = vector<vvi>;
using vl = vector<ll>;
using vvl = vector<vl>;
using vvvl = vector<vvl>;
using vpii = vector<pii>;
using vpll = vector<pll>;
using vs = vector<string>;
template<class T> using pq = priority_queue<T, vector<T>, greater<T>>;
#define overload4(_1, _2, _3, _4, name, ...) name
#define overload3(a,b,c,name,...) name
#define rep1(n) for (ll UNUSED_NUMBER = 0; UNUSED_NUMBER < (n); ++UNUSED_NUMBER)
#define rep2(i, n) for (ll i = 0; i < (n); ++i)
#define rep3(i, a, b) for (ll i = (a); i < (b); ++i)
#define rep4(i, a, b, c) for (ll i = (a); i < (b); i += (c))
#define rep(...) overload4(__VA_ARGS__, rep4, rep3, rep2, rep1)(__VA_ARGS__)
#define rrep1(n) for(ll i = (n) - 1;i >= 0;i--)
#define rrep2(i,n) for(ll i = (n) - 1;i >= 0;i--)
#define rrep3(i,a,b) for(ll i = (b) - 1;i >= (a);i--)
#define rrep4(i,a,b,c) for(ll i = (a) + ((b)-(a)-1) / (c) * (c);i >= (a);i -= c)
#define rrep(...) overload4(__VA_ARGS__, rrep4, rrep3, rrep2, rrep1)(__VA_ARGS__)
#define all1(i) begin(i) , end(i)
#define all2(i,a) begin(i) , begin(i) + a
#define all3(i,a,b) begin(i) + a , begin(i) + b
#define all(...) overload3(__VA_ARGS__, all3, all2, all1)(__VA_ARGS__)
#define sum(...) accumulate(all(__VA_ARGS__),0LL)
template<class T> bool chmin(T &a, const T &b){ if(a > b){ a = b; return 1; } else return 0; }
template<class T> bool chmax(T &a, const T &b){ if(a < b){ a = b; return 1; } else return 0; }
template<class T> auto min(const T& a){ return *min_element(all(a)); }
template<class T> auto max(const T& a){ return *max_element(all(a)); }
template<class... Ts> void in(Ts&... t);
#define INT(...) int __VA_ARGS__; in(__VA_ARGS__)
#define LL(...) ll __VA_ARGS__; in(__VA_ARGS__)
#define STR(...) string __VA_ARGS__; in(__VA_ARGS__)
#define CHR(...) char __VA_ARGS__; in(__VA_ARGS__)
#define DBL(...) double __VA_ARGS__; in(__VA_ARGS__)
#define LD(...) ld __VA_ARGS__; in(__VA_ARGS__)
#define VEC(type, name, size) vector<type> name(size); in(name)
#define VV(type, name, h, w) vector<vector<type>> name(h, vector<type>(w)); in(name)
ll intpow(ll a, ll b){ ll ans = 1; while(b){if(b & 1) ans *= a; a *= a; b /= 2;} return ans;}
ll modpow(ll a, ll b, ll p){ ll ans = 1; a %= p;if(a < 0) a += p;while(b){ if(b & 1) (ans *= a) %= p; (a *= a) %= p; b /= 2; } return ans; }
ll GCD(ll a,ll b) { if(a == 0 || b == 0) return 0; if(a % b == 0) return b; else return GCD(b,a%b);}
ll LCM(ll a,ll b) { if(a == 0) return b; if(b == 0) return a;return a / GCD(a,b) * b;}
namespace IO{
#define VOID(a) decltype(void(a))
struct setting{ setting(){cin.tie(nullptr); ios::sync_with_stdio(false);fixed(cout); cout.precision(30);}} setting;
template<int I> struct P : P<I-1>{};
template<> struct P<0>{};
template<class T> void i(T& t){ i(t, P<3>{}); }
void i(vector<bool>::reference t, P<3>){ int a; i(a); t = a; }
template<class T> auto i(T& t, P<2>) -> VOID(cin >> t){ cin >> t; }
template<class T> auto i(T& t, P<1>) -> VOID(begin(t)){ for(auto&& x : t) i(x); }
template<class T, size_t... idx> void ituple(T& t, index_sequence<idx...>){
    in(get<idx>(t)...);}
template<class T> auto i(T& t, P<0>) -> VOID(tuple_size<T>{}){
    ituple(t, make_index_sequence<tuple_size<T>::value>{});}
#undef VOID
}
#define unpack(a) (void)initializer_list<int>{(a, 0)...}
template<class... Ts> void in(Ts&... t){ unpack(IO :: i(t)); }
#undef unpack
static const double PI = 3.1415926535897932;
template <class F> struct REC {
    F f;
    REC(F &&f_) : f(forward<F>(f_)) {}
    template <class... Args> auto operator()(Args &&...args) const { return f(*this, forward<Args>(args)...); }};
//constexpr int mod = 1000000007;
constexpr int mod = 998244353;

#include<atcoder/mincostflow>
using namespace atcoder;

// https://lorent-kyopro.hatenablog.com/entry/2020/07/24/170656
//BEGIN CUT HERE
struct RangeToRange{
  const int n;
  int ptr;
  RangeToRange(int n):n(n),ptr(3*n){}

  // (0n, 2n) : top segtree (to)
  // (2n, 4n) : bottom segtree (from)

  // map [3n, 4n) -> [n, 2n)
  template<typename F>
  void add_edge(int u,int v,F f){
    if(3*n<=u and u<4*n) u-=2*n;
    f(u,v);
  }

  template<typename F>
  void init(F f){
    for(int i=1;i<n;i++){
      int l=(i<<1)|0;
      int r=(i<<1)|1;
      add_edge(0*n+i,0*n+l,f);
      add_edge(0*n+i,0*n+r,f);
      add_edge(2*n+l,2*n+i,f);
      add_edge(2*n+r,2*n+i,f);
    }
  }

  // [l1, r1) -> [l2, r2)
  template<typename F,typename G>
  void add_edge(int l1,int r1,int l2,int r2,F f,G g){
    int k=ptr++;
    for(l1+=n,r1+=n;l1<r1;l1>>=1,r1>>=1){
      if(l1&1) add_edge(2*n+(l1++),k,f);
      if(r1&1) add_edge(2*n+(--r1),k,f);
    }
    for(l2+=n,r2+=n;l2<r2;l2>>=1,r2>>=1){
      if(l2&1) g(k,l2++);
      if(r2&1) g(k,--r2);
    }
  }
  int idx(int v)const{return 1*n+v;}
  size_t size()const{return ptr;}
};
int main() {
    INT(n,m);
    vi u(m),v(m);
    rep(i,m) {
        cin >> u[i] >> v[i];
        --u[i],--v[i];
    }
    RangeToRange R(2*n);
    mcf_graph<int,int> mcf(R.size() + 2 * n + 2);
    auto f0 = [&](int u,int v) {return mcf.add_edge(u,v,1,0);};
    auto f1 = [&](int u,int v) {return mcf.add_edge(u,v,1,2);};
    int st = 8 * n,gl = 8 * n + 1;
    R.init(f0);
    rep(i,n) {
        mcf.add_edge(st,2*n+i,1,0);
        mcf.add_edge(i+3*n,gl,1,0);
        R.add_edge(i,i+1,n,n+i,f0,f1);
        R.add_edge(i,i+1,n+i+1,2*n,f0,f1);
    }
    rep(i,m) {
        mcf.add_edge(2*n+u[i],v[i]+3*n,1,0);
        mcf.add_edge(2*n+v[i],u[i]+3*n,1,0);
    }
    auto fl = mcf.flow(st,gl);
    
    cout << n - fl.second << '\n';
}