#include <bits/stdc++.h>
using namespace std;

#pragma GCC optimize("Ofast")
#pragma GCC optimize("unroll-loops")

#if __has_include(<atcoder/all>)
#include <atcoder/all>
using namespace atcoder;
#endif

#ifdef LOCAL
#include "algo/debug.hpp"
#else
#define debug(...) void(0)
#endif

using ll = long long;
using ld = long double;
constexpr ll linf=(1LL<<60) - 1;
constexpr int inf=(1LL<<30) - 1;
template <class T> using pq = priority_queue<T>; //max
template <class T> using pqg = priority_queue<T, vector<T>, greater<T>>; //min
#define len(obj) int(obj.size())
#define overload4(a,b,c,d,name,...) name
#define overload5(a,b,c,d,e,name,...) name
#define rep1(n) for(ll _=0;_<n;++_)
#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 _=n-1;_>=0;--_)
#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 each1(i,a) for(auto&&i:a)
#define each2(x,y,a) for(auto&&[x,y]:a)
#define each3(x,y,z,a) for(auto&&[x,y,z]:a)
#define each4(w,x,y,z,a) for(auto&&[w,x,y,z]:a)
#define each(...) overload5(__VA_ARGS__,each4,each3,each2,each1)(__VA_ARGS__)
#define all(x) (x).begin(), (x).end()
#define rall(x) (x).rbegin(), (x).rend() 
#define vec(type,name,...) vector<type> name(__VA_ARGS__)
#define vv(type,name,h,...) vector name(h,vector<type>(__VA_ARGS__))
#define vvv(type,name,h,w,...) vector name(h,vector(w,vector<type>(__VA_ARGS__)))
#define vvvv(type, name, h, w, n, ...) vector<vector<vector<vector<type>>>> name(h, vector<vector<vector<type>>>(w, vector<vector<type>>(n, vector<type>(__VA_ARGS__))))
struct Setting{ Setting(){ cin.tie(nullptr)->sync_with_stdio(0); fixed(cout).precision(12); } }Setting;
//-----------------------------------

using mint = modint998244353;
int main() {
    ll N, M; cin >> N >> M;
    if(N < M) {
        cout << 0 << endl;
    }else {
        mint ans = mint(2).pow(N) - 1;
        mint t = 1;
        rep(i, M - 1) {
            t *= mint(N) - i;
            t /= i + 1;
            ans -= t;
        }
        cout << ans.val() << endl;
    }
}