arc/scripts/standalone_cerebro.mpc at master · pps-lab/arc · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
from Compiler.script_utils import output_utils

from Compiler.script_utils.data import data


from Compiler import ml
from Compiler import library

from Compiler.script_utils.audit import rand_smoothing

from Compiler.script_utils import config, timers, input_consistency

from Compiler.script_utils.consistency_cerebro import compute_commitment


class CerebroConfig(config.BaseAuditModel):

    pass

program.options_from_args()
cfg = config.from_program_args(program.args, CerebroConfig)

def compute_sz(input_flat):
    random_point = 34821
    rho = cint(random_point)

    output_sum = input_flat[0]
    output_sum_r = sint(0)

    # main loop
    # @for_range_multithread(n_threads, 1, inputs.length)
    @for_range_opt(1, input_flat.length)
    def _(i):
        output_sum.update(output_sum + (input_flat[i] * rho))
        output_sum_r.update(output_sum_r + (sint(3) * rho)) # assume r = 3 everywhere
        rho.update(rho * random_point)

    return output_sum, output_sum_r

program.use_trunc_pr = cfg.trunc_pr
sfix.round_nearest = cfg.round_nearest
ml.set_n_threads(cfg.n_threads)

train_dataset, _, _ = data._load_dataset_args(cfg.dataset)
n_players = len(train_dataset)

library.start_timer(timer_id=timers.TIMER_INPUT_CONSISTENCY_CHECK)

for player_id in range(0, n_players):
    objects = input_consistency.read_input_format_from_file(player_id)
    print("Player", player_id, "has", len(objects), "objects")
    for object in objects:
        len_items = sum([item["length"] for item in object["items"]])
        print("Computing commitment of length ", len_items)
        print_ln("Computing commitment for player %s with size %s", player_id, len_items)
        input_comm = Array(len_items, sint)

        # Should load input from secret shares for correctness,
        # we dont to make it faster (since we would have already loaded the data).
        input_comm.assign_all(2)

        output_sum, output_sum_r = compute_sz(input_comm)

        compute_commitment(output_sum, output_sum_r)

library.stop_timer(timer_id=timers.TIMER_INPUT_CONSISTENCY_CHECK)

print_ln("Done computing commitments!")